Off-Label Use Can Boost Sales. It Can Also Break Forecasts.

The Short Answer

In November 2013, Johnson & Johnson agreed to pay more than $2.2 billion to resolve criminal and civil claims that its Janssen unit marketed the antipsychotic Risperdal for dementia, autism, and other uses the FDA never approved.^[1] A decade later, Novo Nordisk raised its full-year sales guidance three times in twelve months because physicians and patients were using Ozempic, a type 2 diabetes drug, for weight loss faster than the company could manufacture it.^[2][3] Both events trace back to the same regulatory fact: a drug’s FDA label covers only the uses the agency reviewed. Everything else — a doctor’s discretion to prescribe beyond that label — is off-label use, and it is both common and commercially unpredictable.

Five numbers frame the rest of this analysis:

• Roughly 21% to 38% of U.S. outpatient prescriptions are for off-label indications, depending on the dataset and time period measured.^[4][5]
• Off-label promotion settlements against drugmakers have totaled more than $9 billion across eight major cases since 2004, based on publicly disclosed Department of Justice figures.^[1][6][7][8]
• A single intravitreal injection of Lucentis cost about $1,950 in 2011, versus about $50 for the off-label alternative, Avastin — a $1,900 gap that reshaped a $2 billion Medicare Part B market.^[9]
• Novo Nordisk’s 2023 sales guidance moved from 13%–19% growth to 32%–38% growth over two revisions, driven substantially by off-label Ozempic demand.^[2][3]
• Independent researchers estimate pharmaceutical commercial forecasts can miss by as much as 40%, and off-label use is one of the variables standard forecasting models are least equipped to capture.^[10]

How Common Is Off-Label Prescribing, Really?

The Federal Baseline: 21% to 38% of All U.S. Prescriptions

The most-cited estimate comes from a 2006 study of 160 commonly prescribed drugs, which found that 21% of prescriptions were for off-label uses.^[4] A separate analysis of National Ambulatory Medical Care Survey data found the off-label prescribing rate rose from 29.9% in 1993 to 38.3% in 2008.^[5] A more recent cross-sectional study using 2016–2021 Medical Expenditure Panel Survey data, covering 1,596,753 prescriptions among 46,770 patients, found that 75% of prescriptions matched a labeled indication while 25% did not — and that only 33% of all prescriptions were both indicated and free of a documented contraindication.^[11] The Congressional Research Service puts the credible range at 12% to 38% of office-based prescriptions, noting that estimates vary by dataset, drug class, and how off-label use is defined.^[12]

Off-Label Use by Drug Class: Cardiovascular Highs, Diabetes Lows

Prevalence varies sharply by therapeutic category. A review in JACC: Basic to Translational Science found off-label use as high as 46% of prescriptions for cardiovascular medications excluding antihyperlipidemics and antihypertensives, versus as low as 1% for diabetes therapies.^[13] Among psychiatric prescribers specifically, one national sample found a 12.9% off-label rate, with the two most common patterns being citalopram prescribed for manic-depressive psychosis and trazodone prescribed for primary insomnia — both established, low-risk off-label patterns rather than aggressive promotion.^[14] Cancer drugs sit at the high end: 41% of oncology drug utilization between 1997 and 2007 was off-label, and researchers classified 29% of that off-label use as clinically appropriate versus 17% as inappropriate.^[15]

Cancer Care Is Different: Compendia Turn Off-Label Into Reimbursable

Off-label oncology prescribing has its own reimbursement infrastructure that other therapeutic areas lack. Under the Social Security Act and CMS’s Medicare Benefit Policy Manual, an off-label cancer drug use is considered “medically accepted” — and therefore payable under Medicare Part B — if it appears in one of a small set of government-recognized compendia, chiefly the National Comprehensive Cancer Network (NCCN) Drugs & Biologics Compendium.^[16][17] CMS formally added the NCCN Compendium to that list, and UnitedHealthcare separately adopted it as the reference for approving off-label chemotherapy claims in outpatient settings.^[16][17]

How the NCCN and CMS Compendium Pathway Works

The mechanism works in three steps. First, a Medicare contractor or private payer checks whether the FDA label covers the use; if it does, standard coverage applies. Second, if the use is off-label, the payer checks whether NCCN, DrugDex, AHFS-DI, or another CMS-recognized compendium lists it with adequate supporting evidence.^[18] Third, if no compendium supports the use, the payer may still review peer-reviewed literature submitted by the prescriber, though this route is slower and less predictable.^[18] Because private payers frequently mirror Medicare’s compendium logic — one unpublished survey of 60 health plans covering more than 90 million members found the majority using a combination of DrugPoints, AHFS-DI, and NCCN compendia — a positive compendium listing can function as a de facto second approval pathway that expands a drug’s addressable, reimbursable market well beyond its FDA label.^[17]

Four Pathways From Off-Label Prescribing to Company Revenue

Off-label revenue does not arrive through a single mechanism. Based on the cases examined in this analysis, off-label use reaches a company’s top line through one of four distinct pathways, each with different legal exposure and forecasting implications. This framework is an original organizing structure for this article, not an established industry classification.

1. Compendia-supported reimbursement. An oncology or other specialty use lacks FDA approval but is listed in NCCN or another recognized compendium, so Medicare and most private payers reimburse it as a matter of policy.^[16][18]
2. Physician-initiated, unpromoted uptake. Clinicians adopt an off-label use based on published evidence or clinical experience, with no manufacturer marketing involved — the pattern behind Avastin’s use in wet age-related macular degeneration (AMD), discussed below.^[19]
3. Illegally promoted off-label use. A manufacturer’s sales force, medical liaisons, or paid publications actively encourage prescribing beyond the label, the pattern underlying the Neurontin, Risperdal, and GSK settlements below.^[1][6][7]
4. Demand-shock off-label use. Media coverage, social platforms, or celebrity endorsement drive patient-initiated requests for an off-label use faster than either the company or regulators anticipated — the pattern behind the Ozempic and Wegovy shortages.^[3][20]

Pathways one and two carry limited legal risk for the manufacturer. Pathway three is the one that has produced nine-figure and ten-figure settlements. Pathway four is legal but has proven just as disruptive to commercial forecasting, because it moves faster than manufacturing capacity or supply contracts can adjust.

What Changed: FDA’s Shifting Rules on Off-Label Communication

United States v. Caronia and the First Amendment Turn

For decades, FDA treated any manufacturer communication promoting an unapproved use as evidence of “misbranding” under the Food, Drug, and Cosmetic Act. That changed in 2012, when the Second Circuit Court of Appeals overturned the conviction of a pharmaceutical sales representative in United States v. Caronia, holding that the FDCA’s misbranding provisions cannot constitutionally criminalize truthful, non-misleading speech promoting an off-label use.^[21]

The Amarin Case and “Truthful, Non-Misleading” Speech

A federal district court extended that logic in 2015 in Amarin Pharma v. FDA, ruling that Caronia’s protection is not limited to a narrow subset of promotional speech, such as responses to unsolicited physician questions or statements from non-sales personnel — the reasoning, the court found, applies across the board to truthful, non-misleading promotional speech generally.^[21] FDA has continued to argue in subsequent litigation, including the 2023 First Circuit decision in United States v. Facteau, that promotional speech can still serve as evidence of an unapproved intended use even when the speech itself is not false.^[22] The result is a body of case law that protects truthful off-label speech in principle while leaving companies without a bright-line safe harbor in practice.

FDA’s January 2025 SIUU Final Guidance

On January 7, 2025, FDA finalized its guidance on what it calls “scientific information on unapproved uses” (SIUU) communications — the agency’s preferred term for what the rest of the industry calls off-label information.^[22] The SIUU Final Guidance updated an October 2023 draft and describes how manufacturers may proactively share scientific information about unapproved uses through reprints, clinical practice guidelines, reference texts, and digital clinical practice resources, provided the communication is truthful and non-misleading.^[22] FDA released the guidance alongside an addendum to its 2017 memorandum on First Amendment considerations, reaffirming its legal position even as it granted companies more room to operate.^[22]

What Companies Can and Cannot Say to Physicians Now

• Manufacturers may proactively distribute SIUU communications directly to any health care provider, not only to payers evaluating coverage, a broadening from earlier FDA policy.^[21]
• Communications must be presented separately from promotional messaging about the approved indication and must disclose that the use discussed is not FDA-approved.
• The underlying science must meet FDA’s bar for truthful, non-misleading content; a favorable court ruling on speech does not immunize a company against enforcement if the specific claims are unsupported or misleading.^[22]
• Guidance documents are not binding law. They describe FDA’s current enforcement thinking, and the agency has changed its position on off-label communication multiple times since 2011.^[23]

When Off-Label Use Boosted Sales — And Then Broke the Law

Neurontin: An Epilepsy Drug That Became a Pain and Psychiatric Blockbuster

Neurontin (gabapentin) was approved as a supplemental epilepsy treatment. Between 1996 and 2000, sales representatives from Warner-Lambert, which Pfizer later acquired, promoted the drug to physicians for migraines, bipolar disorder, ADHD, and various pain conditions — none of which were FDA-approved indications at the time.^[24] The case, Franklin v. Parke-Davis, became the first instance in which a federal court recognized that off-label promotion could trigger False Claims Act liability by causing Medicaid to pay for non-reimbursable prescriptions.^[24]

The Settlement Arithmetic: $430 Million, Then $325 Million

In May 2004, Pfizer’s Warner-Lambert unit pleaded guilty to two felony counts of misbranding and paid $430 million, including a $240 million criminal fine, to resolve the federal and state investigation.^[6] A decade later, in 2014, Pfizer agreed to pay a further $325 million to resolve third-party payer claims covering the period from December 2002 through August 2008, the years when Neurontin prescriptions surged.^[7] Together, the two off-label-promotion settlements total roughly $755 million. (A separate $190 million settlement Pfizer reached the same year involved allegations that it kept generic gabapentin off the market, an antitrust matter distinct from off-label promotion, and should not be added to that off-label total.^[7])

Risperdal: Marketing to Dementia Patients Outside the Label

Risperdal was approved by FDA to treat schizophrenia. Janssen Pharmaceuticals, a Johnson & Johnson subsidiary, promoted it for anxiety, agitation, depression, hostility, and confusion in elderly dementia patients — a population the drug was never studied or approved for, and one federal investigators said was known to carry an increased diabetes risk from the drug that was not adequately disclosed.^[1][25]

Whistleblowers and the $2.2 Billion Total

In November 2013, Johnson & Johnson and Janssen agreed to pay more than $2.2 billion — $485 million in criminal fines and forfeiture, plus $1.72 billion in civil penalties — to resolve allegations covering Risperdal, the related antipsychotic Invega, and the heart failure drug Natrecor, along with kickbacks paid to the long-term care pharmacy provider Omnicare.^[1] At the time, the Justice Department called it one of the largest health care fraud settlements in U.S. history, and attorneys involved described it as the largest settlement ever tied to a single drug.^[25] Six whistleblowers, several of them former sales representatives, split more than $167 million of the recovery.^[25]

Paxil, Wellbutrin, and Avandia: GSK’s $3 Billion Reckoning

GlaxoSmithKline’s 2012 settlement remains the largest health care fraud resolution in U.S. history at $3 billion.^[8] The government’s allegations spanned three separate off-label and safety failures: marketing the antidepressant Paxil for pediatric use it was never approved for; marketing the antidepressant Wellbutrin for weight loss, sexual dysfunction, and substance abuse, and at higher-than-approved doses; and failing to report safety data to FDA about cardiovascular risk with the diabetes drug Avandia.^[8] GSK pleaded guilty to a three-count criminal information and paid $1 billion in criminal fines and forfeiture, plus $2 billion in civil penalties.^[8]

The Sales Figures Behind the Settlement

According to IMS Health data cited at the time, the three drugs generated substantial revenue across the years the settlement covered: Avandia brought in $10.4 billion, Paxil $11.6 billion, and Wellbutrin $5.9 billion.^[26] A spokesperson for the whistleblower advocacy group Taxpayers Against Fraud noted that a $3 billion settlement for half a dozen drugs over a decade “can be rationalized as the cost of doing business.”^[26] Prosecutors said at the time they could not determine precisely how much of GSK’s profit came specifically from the off-label marketing, though they suggested it plausibly exceeded the settlement amount.^[26]

A Pattern, Not an Outlier: Aranesp, OxyContin, Botox, Seroquel, Abilify

The Neurontin, Risperdal, and GSK cases are the largest, but they are not isolated. Amgen paid $762 million in 2012 over off-label promotion and kickback allegations tied to the anemia drug Aranesp.^[27] Purdue Pharma paid $601 million in 2007 over off-label promotion of OxyContin.^[27] Allergan paid $600 million in 2010 over off-label promotion of Botox.^[27] AstraZeneca paid $520 million in 2010 over off-label promotion and kickbacks involving Seroquel.^[27] Bristol-Myers Squibb paid $515 million in 2007 over off-label promotion, kickbacks, and Medicare fraud tied to Abilify.^[27] Across just these eight cases — Neurontin, Risperdal, Paxil/Wellbutrin/Avandia, Aranesp, OxyContin, Botox, Seroquel, and Abilify — disclosed settlement totals exceed $9 billion.

Company	Product(s)	Year	Settlement	Primary Allegation
Pfizer / Warner-Lambert	Neurontin	2004 & 2014	$430M + $325M	Off-label promotion (pain, psychiatric, migraine)
Johnson & Johnson / Janssen	Risperdal, Invega, Natrecor	2013	$2.2B	Off-label promotion, kickbacks
GlaxoSmithKline	Paxil, Wellbutrin, Avandia	2012	$3B	Off-label promotion, failure to report safety data
Amgen	Aranesp	2012	$762M	Off-label promotion, kickbacks
Purdue Pharma	OxyContin	2007	$601M	Off-label promotion
Allergan	Botox	2010	$600M	Off-label promotion
AstraZeneca	Seroquel	2010	$520M	Off-label promotion, kickbacks
Bristol-Myers Squibb	Abilify	2007	$515M	Off-label promotion, kickbacks, Medicare fraud

Source: Department of Justice and state attorney general press releases, compiled via Wikipedia’s List of largest pharmaceutical settlements.^[27]

When Off-Label Use Boosted Sales — Without Breaking Any Law

Avastin vs. Lucentis: A Company Competing With Its Own Off-Label Use

Genentech developed two closely related anti-VEGF drugs from the same underlying science. Avastin was approved by FDA in 2004 for metastatic colorectal cancer. Lucentis, derived from a similar protein but engineered specifically for intraocular use, was approved in June 2006 for wet AMD, a leading cause of blindness in older adults.^[28] By 2005 — a year before Lucentis reached the market, and while Lucentis was still in Genentech-sponsored trials — ophthalmologists had already begun injecting small, off-label doses of Avastin into patients’ eyes, reasoning that its similarity to Lucentis and its availability made it worth trying.^[29]

The National Eye Institute funded the Comparison of AMD Treatments Trials (CATT) specifically because so many patients were already receiving the unapproved option: “Over 250,000 patients are treated each year for AMD, and a substantial number of them receive Avastin,” NEI director Paul Sieving said at the time, adding that the lack of efficacy data for that off-label use created an obligation to study it directly.^[29] The first-year CATT results, published in 2011, found Avastin equally effective as Lucentis for treating wet AMD.^[29]

The CATT Trial Numbers: $1,950 vs. $50 Per Injection

The price differential explains why the equivalence finding mattered commercially. In 2011, a single Lucentis injection cost about $1,950, compared with about $50 for the compounded off-label dose of Avastin — a roughly 39-to-1 cost ratio for what CATT found to be a clinically comparable treatment.^[9] Combined Medicare Part B spending on the two drugs approached $2 billion in 2010 alone.^[9] A U.S. Department of Health and Human Services Office of Inspector General survey found that most ophthalmologists who chose Avastin over Lucentis cited the cost difference as their primary reason.^[9] Following the CATT results, prescribing share measurably shifted away from the more expensive, on-label Lucentis and toward the off-label Avastin.^[30] Genentech, in effect, had built a forecasting problem into its own product portfolio: it profited on Avastin’s oncology sales while its investment in the FDA-approved, ophthalmology-specific Lucentis faced a self-inflicted, off-label substitute.

Ozempic and Wegovy: Guidance Raised Three Times in a Year

Semaglutide, the active ingredient in Novo Nordisk’s Ozempic and Wegovy, was approved for type 2 diabetes management under the Ozempic brand. A 2021 clinical trial demonstrated its effectiveness for weight loss, and FDA approved the higher-dose formulation as Wegovy specifically for chronic weight management that same year.^[31] But patients seeking weight loss began requesting the diabetes-branded Ozempic off-label, partly because insurance more readily covered a diabetes diagnosis and partly because Wegovy itself quickly became supply-constrained.^[32]

From 13%–19% Growth to 32%–38% Growth

The forecasting effect shows up directly in Novo Nordisk’s own guidance revisions. In April 2023, the company raised its full-year sales growth guidance from a previous range of 13%–19% to 24%–30%, citing higher demand for both Wegovy and off-label Ozempic use for weight loss.^[32] By October 2023, Novo raised guidance again, to 32%–38% sales growth, again citing huge demand for Ozempic, “which is often used off-label to treat obesity.”^[3] GlobalData had separately forecast 23% Ozempic sales growth for 2023 in a May 2023 analysis that explicitly named the drug’s off-label weight-loss application as a growth driver.^[2] Novo ultimately reported $33.7 billion in full-year 2023 sales, up 36% at constant exchange rates and ahead of analyst expectations, with its obesity and diabetes care division — $31.2 billion of that total — posting 154% growth in obesity care sales and 52% growth in GLP-1 diabetes medication sales.^[33] The company’s market value rose from roughly $160 billion at the start of 2021 to roughly $480 billion by January 2024, making it briefly Europe’s most valuable public company.^[33]

The same demand that inflated Novo’s guidance also produced shortages. Australia’s Therapeutic Goods Administration stated that Ozempic would be unavailable in the country until the end of March 2023 and attributed the shortfall in part to off-label prescribing.^[34] Asked directly about off-label use aggravating the shortage, a Novo Nordisk spokesperson said the company does not promote, suggest, or encourage off-label use, and that treatment decisions remain up to each physician’s clinical discretion.^[34] Novo told analysts in late 2023 that supply constraints and related shortage notifications would likely persist into 2024, and reporting in 2024 indicated constraints on lower starting doses continuing into 2025.^[35] A June 2023 survey of 3,420 U.S. adults found that 58% were at least somewhat concerned about Ozempic shortages caused by off-label use, while public opinion on whether physicians should have unrestricted discretion to prescribe off-label was closely split, at 46.3% in favor versus 37.7% wanting federal regulation of the practice.^[36]

Why Off-Label Demand Breaks Forecast Models

Standard Epidemiology-Based Models Are Built for the Label, Not the Market

Most pharmaceutical commercial forecasts start from an epidemiology-based patient funnel: total prevalence of the approved condition, diagnosis rates, treatment rates, and expected market share within the treated population.^[37] That structure works well for a drug used exactly as labeled. It has no natural place for demand that originates outside the labeled population — a diabetes-drug funnel simply does not include patients who want the drug for weight loss, and an oncology funnel built around FDA-approved lines of therapy does not automatically include every compendium-supported off-label regimen a physician might choose.^[38]

The 40 Percent Forecast Error Problem

Forecasting specialists at Evaluate note that pharmaceutical sales forecasts can be off by as much as 40%, often because of poorly specified assumptions rather than poor data collection.^[10]

“Pharmaceutical forecasts can be out by as much as 40 percent,” according to research cited by forecasting analytics firm Evaluate, which attributes the gap to weak model transparency and unjustified assumptions rather than a shortage of underlying data.^[10]

Standard mitigation techniques — Monte Carlo simulation and structured sensitivity analysis — are designed to quantify uncertainty around known variables like price, market share, and rate of uptake.^[39] Off-label demand is harder to model this way because it is not merely uncertain within a known range; in cases like Ozempic, it changes the size and definition of the addressable market itself, mid-forecast-cycle.^[32]

Supply Chain Consequences: Shortages, Rationing, and Diverted Care

When off-label demand outruns a forecast, the downstream effect is not merely a spreadsheet miss — it becomes a supply chain and patient-access problem. Novo Nordisk restricted starter doses of Wegovy to preserve supply for existing patients while working to add a second contract manufacturer.^[32] Reports from multiple countries described diabetic patients struggling to access Ozempic because available supply was being diverted to patients using it for weight loss.^[32] The American Society of Health-System Pharmacists tracked shortages across several Ozempic dose strengths into early 2023, and FDA separately confirmed supply constraints on specific formulations extending into that same period.^[34]

What the Law Actually Prohibits — and Permits

Physicians Can Prescribe Off-Label. Companies Cannot Market It.

This distinction sits at the center of every case in this analysis. Once FDA approves a drug for any indication, a licensed physician may lawfully prescribe it for any use they judge appropriate, including uses far outside the FDA-reviewed label — off-label prescribing itself has never been illegal.^[13] What the FDCA restricts is manufacturer promotion of unapproved uses, historically treated as evidence that a drug was “misbranded” because its labeling did not include adequate directions for the use being promoted.^[24] The Caronia and Amarin rulings narrowed that restriction for truthful, non-misleading manufacturer speech, but they did not eliminate FDA’s authority to act against false or misleading off-label claims, and they did not touch the underlying rule that physicians retain prescribing discretion regardless of a manufacturer’s conduct.^[21]

The False Claims Act Connection

Franklin v. Parke-Davis established the mechanism that turned off-label promotion into billion-dollar exposure: when a manufacturer’s off-label marketing causes a physician to prescribe a drug for a use that a government payer like Medicaid does not cover, and that prescription is then billed to the government, the resulting claim can be “false” under the False Claims Act — even though the prescription itself was legal.^[24] That theory, first validated in the Neurontin litigation, is the legal foundation underneath every subsequent nine- and ten-figure off-label settlement described in this article.^[24]

Where AI-Generated Drug Information Fits Into the Off-Label Picture

How AI Chatbots Handle Questions About Unapproved Uses

Patients and, increasingly, clinicians ask general-purpose AI chatbots questions about drugs before or instead of consulting a prescribing label. A drug’s off-label reputation — built through media coverage, patient forums, and social platforms, as with Ozempic — can shape what an AI system says even when that reputation has nothing to do with the FDA-approved indication.^[20] Because large language models are trained on broad internet text rather than curated, current prescribing information, an AI answer about a drug’s uses can blend approved indications, compendium-supported off-label uses, informal off-label patterns, and outdated information without clearly distinguishing among them unless it is specifically designed to do so.

Documented Accuracy and Hallucination Rates in Medical AI

Independent research on AI medical accuracy shows why that blending matters. A 2024 study in the Journal of Medical Internet Research found that ChatGPT and Bard generated hallucinated, non-existent references in 28.6% to 91.3% of cases when asked to support systematic review claims, depending on the specific query set tested.^[40] A related evaluation of reference-checking tools found that ChatGPT and Bing reached “critical” hallucination levels, while research-specific tools like Elicit and SciSpace showed negligible hallucination on the same task.^[41] In a separate benchmark testing large language models specifically on drug-drug interaction questions, accuracy stayed below 26% across every model tested, with the best-performing systems — GPT-4-turbo and GPT-3.5-turbo — still correctly identifying interactions only about a quarter of the time.^[42]

What a Discrepancy Analysis Looks Like in Practice

A methodologically sound comparison of AI-generated drug information against authoritative sources specifies, at minimum: the AI platform and model version tested; the exact prompts used; the date and time of testing, since model outputs change with updates; whether web search or retrieval augmentation was enabled; the reference source used for verification, such as the current FDA label or a specific compendium version; and how discrepancies were classified and their severity scored.^[43] Tools purpose-built to run this kind of comparison at scale — DrugChatter’s monitoring of AI-generated drug answers against current prescribing information is one example — give medical affairs and pharmacovigilance teams a repeatable way to see where a model’s off-label framing diverges from the label, rather than relying on ad hoc spot checks. Any such comparison is only as good as its documented methodology and should never be presented as proof that an AI answer is medically authoritative simply because it reads fluently.

Why This Matters More as Patients Ask AI Before Asking a Doctor

The Ozempic case illustrates the stakes: public demand for an off-label use was strong enough to move a $480 billion company’s manufacturing priorities.^[33] If AI systems become a common first stop for patients researching that kind of use, the accuracy of an AI answer about what is and is not FDA-approved becomes a pharmacovigilance and medical-information concern in its own right, independent of anything a manufacturer says or does.

What This Means for Medical Affairs

Medical affairs teams remain the primary legal channel for discussing off-label information with physicians, particularly for unsolicited questions. FDA’s January 2025 SIUU Final Guidance gives medical affairs more room to proactively share truthful, non-misleading scientific information about unapproved uses, but only through defined formats — reprints, clinical practice guidelines, reference texts, and digital clinical practice resources — and only when clearly separated from promotional messaging.^[22]

Handling Unsolicited Off-Label Questions from HCPs

The FDA’s long-standing draft guidance on responding to unsolicited requests, first proposed in 2011 and still the operative framework alongside the 2025 SIUU guidance, distinguishes between a physician’s specific, individualized question and a broader promotional message.^[17] Medical affairs teams should document the unsolicited nature of each request, respond with balanced information that includes any relevant safety findings, and route any pattern of repeated requests about a specific off-label use to the medical information and compliance functions for visibility.

What This Means for Pharmacovigilance

Standard pharmacovigilance infrastructure was not built to track off-label use. The FDA’s Adverse Event Reporting System (FAERS) does not specifically flag whether a reported adverse event occurred during an approved or off-label use, and large-scale observational safety surveillance programs have historically shared that same limitation.^[43]

The Off-Label Signal Detection Gap

This creates a structural blind spot: if an adverse event pattern is specific to an off-label population — different dosing, different comorbidities, different concomitant medications — a signal detection system built around the labeled population may under-detect it. Researchers have built automated, text-mining approaches specifically to identify off-label use from clinical notes because routine surveillance systems do not capture it natively.^[44] Pharmacovigilance teams working with drugs known to carry meaningful off-label volume, per the prevalence data above, should treat that volume as a specific input into signal detection planning rather than an edge case.

What This Means for Brand and Commercial Teams

The Avastin/Lucentis and Ozempic/Wegovy cases show that off-label demand can move in either direction relative to a brand strategy: it can cannibalize an on-label, higher-margin product, as it did for Lucentis, or it can outrun supply and inflate near-term revenue in ways that create later access and reputational risk, as with Ozempic.^[9][32] Commercial and brand teams should treat unpromoted off-label volume as a market signal worth quantifying — not to promote it, which remains legally restricted, but to size the manufacturing, supply, and access planning implications before a shortage or a competitor’s compendium listing forces a reactive response.

What This Means for Regulatory Affairs

The legal landscape governing off-label communication has shifted materially in the past three years, from the October 2023 SIUU draft guidance to the January 2025 SIUU Final Guidance and its accompanying First Amendment addendum, against a backdrop of case law — Caronia, Amarin, and Facteau — that has not fully settled.^[21][22] Regulatory affairs teams should treat FDA guidance documents as the agency’s current enforcement posture rather than fixed law, since FDA has revised its stated approach to off-label communication multiple times since its first attempt at unsolicited-request guidance in 2011.^[17]

Key Takeaways

• Off-label prescribing affects an estimated 21% to 38% of U.S. outpatient prescriptions, varying widely by drug class, from roughly 1% for diabetes therapies to 46% for certain cardiovascular medications.^[4][5][13]
• Eight major off-label promotion settlements since 2004 — covering Neurontin, Risperdal, Paxil/Wellbutrin/Avandia, Aranesp, OxyContin, Botox, Seroquel, and Abilify — total more than $9 billion in disclosed penalties.^{[1][6][7][8][27]}
• Physicians can lawfully prescribe off-label; manufacturers cannot lawfully promote it, and the False Claims Act theory validated in the Neurontin litigation is the legal basis for most subsequent settlements.^[24]
• FDA’s January 2025 SIUU Final Guidance expanded, but did not eliminate, restrictions on proactive manufacturer communication about unapproved uses.^[22]
• Avastin’s off-label use in wet AMD, priced at roughly $50 per injection versus $1,950 for the FDA-approved Lucentis, shifted a $2 billion Medicare Part B market after the CATT trial confirmed clinical equivalence.^[9][29]
• Off-label demand for Ozempic pushed Novo Nordisk to raise its sales guidance three times in 2023, ending the year with $33.7 billion in sales, while simultaneously causing supply shortages that persisted into 2025.^[3][33][35]
• Documented AI hallucination rates on medical reference and drug-interaction questions range from under 26% accuracy on drug-drug interactions to over 90% reference hallucination on some systematic-review tasks, underscoring the need for methodical, sourced verification of AI-generated drug information rather than face-value trust.^[40][42]

FAQ

What does “off-label” actually mean? It means a drug is being prescribed for an indication, patient population, dosage, or route of administration that FDA has not reviewed and approved for that specific drug.^[13]Is it illegal for a doctor to prescribe a medication off-label? No. Physician prescribing discretion covers off-label use, and courts and regulators have consistently distinguished the physician’s lawful prescribing decision from a manufacturer’s separate, more restricted right to promote that use.^[13][21]Can a drug company ever legally talk about off-label uses? Yes, within limits. Following Caronia, Amarin, and FDA’s January 2025 SIUU Final Guidance, companies may share truthful, non-misleading scientific information about unapproved uses through specific formats like reprints and clinical practice guidelines, but not as unrestricted promotional messaging.^[21][22]Why did Avastin become the default treatment for wet AMD instead of the FDA-approved Lucentis? The CATT trial found the two drugs clinically equivalent for wet AMD, and Avastin cost about $50 per injection versus roughly $1,950 for Lucentis, so cost drove a documented prescribing shift toward the off-label option after the trial results were published.^[9][29][30]How much have off-label marketing settlements cost the pharmaceutical industry? Across eight of the largest publicly disclosed cases since 2004, settlements total more than $9 billion, led by GlaxoSmithKline’s $3 billion settlement and Johnson & Johnson’s $2.2 billion settlement.^[1][8][27]Does Medicare ever pay for off-label drug use? Yes, primarily in oncology. If an off-label cancer drug use appears in a CMS-recognized compendium such as the NCCN Drugs & Biologics Compendium, Medicare Part B generally covers it as a “medically accepted” indication.^[16][18]Why did Novo Nordisk keep raising its Ozempic and Wegovy sales guidance in 2023? Off-label demand for Ozempic as a weight-loss treatment, combined with strong on-label Wegovy demand, repeatedly exceeded the company’s own forecasts, pushing guidance from 13%–19% growth to 32%–38% growth over two revisions that year.^[2][3][32]Can AI chatbots reliably tell patients whether a drug’s use is FDA-approved or off-label? Not consistently. Documented studies show AI accuracy on drug-drug interaction questions below 26% and reference hallucination rates as high as 91.3% on certain medical literature tasks, which is why systematic, sourced comparison against current prescribing information matters more than a single fluent-sounding answer.^[40][42]What’s the difference between what the CATT trial showed and what an FDA approval requires? CATT demonstrated clinical equivalence between Avastin and Lucentis for one specific condition, wet AMD, under a specific trial protocol; it did not constitute, and was never intended to constitute, an FDA approval of Avastin for that use, which remains off-label to this day.^[29]How should a company communicate off-label information to physicians under the 2025 FDA guidance? Through the formats FDA’s SIUU Final Guidance describes — reprints, clinical practice guidelines, reference texts, and digital clinical practice resources — kept clearly separate from promotional material about the approved indication, with the unapproved status disclosed and the underlying science held to a truthful, non-misleading standard.^[22]

References

1. U.S. Department of Justice. (2013, November 4). Johnson & Johnson to pay more than $2.2 billion to resolve criminal and civil investigations [Press release]. https://www.justice.gov/archives/opa/pr/johnson-johnson-pay-more-22-billion-resolve-criminal-and-civil-investigations
2. GlobalData. (2023, May 15). Novo Nordisk’s blockbuster Ozempic sales to surge by 23% in 2023, forecasts GlobalData. https://www.globaldata.com/media/pharma/novo-nordisks-blockbuster-ozempic-sales-surge-23-2023-forecasts-globaldata/
3. Fierce Pharma. (2023, October 13). High-flying Novo Nordisk again dials up its 2023 sales and profit expectations. https://www.fiercepharma.com/pharma/novo-nordisk-dials-sales-guidance-record-high-again-soaring-ozempic-wegovy
4. Off-Label Use vs Off-Label Marketing of Drugs: Part 1. (2023). JACC: Basic to Translational Science. https://www.jacc.org/doi/10.1016/j.jacbts.2022.12.011
5. Bradford, W. D., et al., as cited in: Off-Label Use Vs Off-Label Marketing of Drugs, Part 1. JACC: Basic to Translational Science. https://www.jacc.org/doi/10.1016/j.jacbts.2022.12.011
6. Office of the Attorney General, State of California. (2004, May 13). Attorney General Lockyer announces $430 million settlement with Pfizer, Warner-Lambert [Press release]. https://oag.ca.gov/node/29679
7. Fierce Pharma. (2014, June 2). Pfizer adds another $325M to Neurontin settlement tally. Total? $945M. https://www.fiercepharma.com/sales-and-marketing/pfizer-adds-another-325m-to-neurontin-settlement-tally-total-945m
8. Office of the Attorney General, State of Alaska. (2012, August 28). GlaxoSmithKline to pay $3 billion to settle drug marketing and pricing claims [Press release]. https://law.alaska.gov/press/releases/2012/082812-GSKsettlement.html
9. Medscape. (2012, April 27). For AMD, off-label Avastin cheaper than on-label Lucentis. https://www.medscape.com/viewarticle/762824
10. Evaluate. (2025). What makes a good (and bad) pharmaceutical forecast model. https://www.evaluate.com/blog/what-makes-a-good-and-bad-pharmaceutical-forecast-model/
11. Prevalence and relationship with health of off-label and contraindicated drug use in the United States: a cross-sectional study. (2026). PMC. https://pmc.ncbi.nlm.nih.gov/articles/PMC11892052/
12. Congressional Research Service. (2021, February 23). Off-label use of prescription drugs (R45792). https://www.congress.gov/crs_external_products/R/PDF/R45792/R45792.3.pdf
13. Off-Label Use Vs Off-Label Marketing of Drugs, Part 1. (2023). ScienceDirect / JACC: Basic to Translational Science. https://www.sciencedirect.com/science/article/pii/S2452302X23000116
14. Patterns and predictors of off-label prescription of psychiatric drugs. PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6053129/
15. Smieliauskas, F., et al., as cited in: Prevalence and relationship with health of off-label and contraindicated drug use in the United States. https://pmc.ncbi.nlm.nih.gov/articles/PMC11892052/
16. National Comprehensive Cancer Network. CMS recognizes the NCCN Drugs & Biologics Compendium [News release]. https://www.nccn.org/home/news/newsdetails?NewsID=157
17. Oncology Times. (2008). Compendia decisions impact off-label coverage. https://journals.lww.com/oncology-times/fulltext/2008/06250/compendia_decisions_impact_off_label_coverage.10.aspx
18. Noridian Healthcare Solutions. Determination of approved and accepted off-label drug indications. https://med.noridianmedicare.com/web/jfb/topics/drugs-biologicals-injections/determination-of-approved-and-accepted-off-label-drug-indications
19. National Eye Institute. (2011). NIH study finds Avastin and Lucentis are equally effective in treating age-related macular degeneration. https://www.nei.nih.gov/research-and-training/research-news/nih-study-finds-avastin-and-lucentis-are-equally-effective-treating-age-related-macular-degeneration
20. Cambridge Core. The prevalence and consequences of support for off-label Ozempic prescriptions. Health Economics, Policy and Law. https://www.cambridge.org/core/journals/health-economics-policy-and-law/article/prevalence-and-consequences-of-support-for-offlabel-ozempic-prescriptions/45D1BB9616BBF5937F856CC3E2B2AFC9
21. Arnold & Porter. (2023, October). FDA updates off-label communications guidance. https://www.arnoldporter.com/en/perspectives/advisories/2023/10/fda-updates-off-label-communications-guidance
22. Ropes & Gray LLP. (2025, January 22). FDA finalizes guidance on communication of scientific information on unapproved uses and releases updated First Amendment analysis. https://www.ropesgray.com/en/insights/alerts/2025/01/fda-finalizes-guidance-on-communication-of-scientific-information
23. Drug & Device Law Blog. (2023, November). FDA and off-label communication: Getting closer to truth. https://www.druganddevicelawblog.com/2023/11/fda-and-off-label-communication-getting-closer-to-truth.html
24. Wikipedia. Franklin v. Parke-Davis. https://en.wikipedia.org/wiki/Franklin_v._Parke-Davis
25. Campbell Law Observer / topclassactions.com. (2013). Johnson & Johnson to pay $2.2B in Risperdal misbranding settlement. https://topclassactions.com/lawsuit-settlements/prescription/johnson-johnson-to-pay-2-2b-in-risperdal-misbranding-settlement/
26. TIME. (2012, July 5). Breaking down GlaxoSmithKline’s billion-dollar wrongdoing. https://healthland.time.com/2012/07/05/breaking-down-glaxosmithklines-billion-dollar-wrongdoing/
27. Wikipedia. List of largest pharmaceutical settlements. https://en.wikipedia.org/wiki/List_of_largest_pharmaceutical_settlements
28. Genentech. (2011, April 28). Genentech statement on CATT trial data published in the New England Journal of Medicine. https://www.gene.com/media/statements/ps_042811
29. National Eye Institute. Comparison of AMD Treatments Trials (CATT): Lucentis-Avastin trial. https://www.nei.nih.gov/about/news-and-events/news/comparison-amd-treatments-trials-catt-lucentis-avastin-trial
30. Fierce Pharma. (2011). Impact of CATT study findings evident in significant reported changes in prescribing of Genentech’s Avastin and Lucentis. https://www.fiercepharma.com/pharma/impact-of-catt-study-findings-evident-significant-reported-changes-prescribing-of-genentech
31. pharmaphorum. (2023, November 2). Novo Nordisk shrugs off supply issues with record sales. https://pharmaphorum.com/news/novo-nordisk-shrugs-supply-issues-record-sales
32. BioPharma Dive. (2023, April 13). Novo boosts sales outlook on strong demand for obesity drug. https://www.biopharmadive.com/news/novo-raises-wegovy-ozempic-sales-outlook-obesity/647568/
33. Forbes. (2024, January 31). Ozempic maker Novo Nordisk beats profit forecasts as weight-loss sales soar. https://www.forbes.com/sites/roberthart/2024/01/31/novo-nordisk-beats-profit-forecasts-as-weight-loss-sales-soar/
34. MM+M. (2022, December). Novo Nordisk comments on role of off-label use in Ozempic shortfall. https://www.mmm-online.com/home/channel/novo-nordisk-addresses-role-of-off-label-use-in-ozempic-shortfall/
35. Supply Chain Digital. (2024, August 5). Novo Nordisk’s Ozempic shortage: Smarter forecasting needed. https://supplychaindigital.com/articles/novo-nordisk-ozempic-supplier-faces-unprecedented-demand
36. Cambridge Core. The prevalence and consequences of support for off-label Ozempic prescriptions. https://www.cambridge.org/core/journals/health-economics-policy-and-law/article/prevalence-and-consequences-of-support-for-offlabel-ozempic-prescriptions/45D1BB9616BBF5937F856CC3E2B2AFC9
37. Pharmaceutical Commerce. Forecasting for the pharmaceutical industry. https://www.pharmaceuticalcommerce.com/view/forecasting-for-the-pharmaceutical-industry
38. Foster Rosenblatt. Pharma demand forecasting. https://rlsconsultants.com/pharma-demand-forecasting/
39. Reuters Events Pharma. Understanding uncertainty and risk in pharma forecasting. https://www.reutersevents.com/pharma/commercial/understanding-uncertainty-and-risk-pharma-forecasting
40. Chelli, M., Descamps, J., Lavoué, V., Trojani, C., Azar, M., Deckert, M., Raynier, J. L., Clowez, G., Boileau, P., & Ruetsch-Chelli, C. (2024). Hallucination rates and reference accuracy of ChatGPT and Bard for systematic reviews: Comparative analysis. Journal of Medical Internet Research, 26, e53164. https://pmc.ncbi.nlm.nih.gov/articles/PMC11153973/
41. JMIR Medical Informatics. (2024). Reference hallucination score for medical artificial intelligence chatbots: Development and usability study. https://medinform.jmir.org/2024/1/e54345
42. Reliability of large language model knowledge across brand and generic cancer drug names (OncoRABBITS). (2024). medRxiv. https://www.medrxiv.org/content/10.1101/2024.11.08.24316949.full.pdf
43. FDA. Guidance for industry: Responding to unsolicited requests for off-label information about prescription drugs and medical devices, referenced via FDA guidance document 184871. https://www.fda.gov/media/184871/download
44. Automated Detection of Off-Label Drug Use. PMC. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3929699/