Patent Protection Is Not a Moat: What Vivos, GLP-1 Drugs, and Forty Years of Sleep Apnea Belief Actually Tell Us

Every sleep-apnea investor presentation I have read in the last three years contains the same number: roughly 80% of obstructive sleep apnea cases go undiagnosed. It appears in Vivos Therapeutics' materials (NASDAQ: VVOS) alongside its patent protection claims, in ResMed's, in Inspire's, in decks from companies that no longer exist. The number does a specific job. It converts a clinical problem into a total addressable market, and it does so without anyone having to explain where it came from.

I spent about six weeks reading backwards from that figure — through the patent family, the FDA clearance summaries, the trial that generated it, and the GLP-1 obesity trial that may quietly obsolete the market-sizing logic built on top of it. This is what I found.

The verdict: Vivos's intellectual property position is real and defensible, but it fences a field whose boundaries are being redrawn by injectable weight-loss drugs, and no one — including me — has data on how oral appliance therapy and GLP-1 agonists interact in the same patient.

What I could and couldn't test

I read the issued claims, not just the press releases. I read the 510(k) summaries for the DNA appliance family and the 2023 clearance covering severe OSA. I read SURMOUNT-OSA, the tirzepatide sleep apnea trial, in full, including supplementary tables. I traced the 80% figure and the CPAP-adherence figure back to their originating papers and read those.

What I did not do: see a patient. Run a polysomnogram. Fit an appliance. I have no access to Vivos's unpublished treatment registry, which is where the most interesting data on this question almost certainly lives. My sample size for clinical claims is zero, and everything below is a reading of published evidence and a reading of a market, not a bedside observation. Where I'm inferring, I say so.

Where the number came from

The 80%-undiagnosed figure traces, through a long chain of citations, to work by Terry Young and colleagues on the Wisconsin Sleep Cohort — a study of state employees aged 30 to 60, recruited beginning in 1988, polysomnographed in a lab. The 1993 paper established prevalence. A 1997 follow-up estimated that something like 82% of men and 93% of women with moderate-to-severe sleep-disordered breathing had not received a clinical diagnosis.

That estimate was accurate for what it measured: Wisconsin state employees, in the early 1990s, before sleep medicine had built out its clinical infrastructure, before home sleep testing existed, before primary care physicians routinely asked about snoring. It was a snapshot of a diagnostic system that barely existed yet.

The number is now roughly thirty years old, and the denominator has moved twice. First, obesity prevalence in US adults has climbed substantially since the cohort was recruited, which pushes true prevalence up and makes the undiagnosed fraction potentially larger in absolute terms. Second, diagnostic capacity has expanded enormously — home sleep apnea tests, consumer wearables flagging oxygen desaturation, dentists screening for it — which should push the undiagnosed fraction down. These two forces move in opposite directions, and I could find no recent study that resolves them cleanly at national scale.

So when a deck says 80% undiagnosed, it is not lying. It is quoting a real measurement from a real cohort. It is just quoting it as though nothing has happened since 1997, and using it as the multiplicand in a market-size calculation that the original authors never proposed.

The second load-bearing number

The same archaeology applies to the claim that props up every CPAP-alternative pitch: that CPAP adherence is poor.

The canonical citation is Kribbs et al., 1993 — thirty-five patients, covert monitoring of actual machine use, and the finding that patients dramatically overreported their usage. Objective adherence at the seven-hours-per-night standard was dismal. That paper was genuinely important; it established that self-reported CPAP use is unreliable, which was news.

But n=35, in 1993, on 1993 hardware. The machines were louder. The masks were worse. There was no auto-titration, no heated humidification, no ramp feature, no cloud-connected adherence coaching, no employer or payer program nudging usage. Modern adherence figures from large registry datasets are meaningfully better than the Kribbs-era picture — not good, but not the catastrophe the 1993 study captured.

The belief hardened faster than the evidence updated. "CPAP compliance is around 50%" became something everyone knows, repeated in decks, in dental CE courses, in journalism. It has a source. The source is thinner and older than the belief.

I want to be precise about what I am and am not saying. CPAP adherence remains a real clinical problem. Patients do abandon therapy, and the ones who abandon it are not a small group. But there is a difference between "a substantial minority of patients cannot tolerate this device" and "CPAP has failed," and the second framing is doing commercial work the first framing doesn't support.

The comparison

Here is how the four main approaches actually stack up on named criteria. Figures are drawn from published trials and public pricing; where ranges are wide I've said so rather than picking a flattering midpoint.

The row that should stop an investor is the fourth one. A weekly injection produced AHI reductions in the range of what a good appliance produces, in patients with obesity and moderate-to-severe OSA, and did it as a side effect of treating something else entirely.

The interaction nobody has trialed

Here is where the piece gets genuinely uncertain, and I would rather say that than paper over it.

GLP-1 receptor agonists — tirzepatide, semaglutide, and the pipeline behind them — produce weight loss in the 15-22% range in trial populations. Sleep apnea severity is strongly, though not exclusively, tied to body weight and specifically to fat deposition in the parapharyngeal tissue and tongue. Remove 20% of body mass and you change the airway.

What this does to oral appliance therapy is, as far as I can determine from published literature, unstudied. Several interactions are plausible and none are established:

Fit drift. Oral appliances are fitted to a specific dentition and a specific soft-tissue envelope. A patient who loses forty-five pounds over a year of treatment has a different airway and, in some cases, meaningfully different facial soft tissue. Whether a device titrated at baseline remains correctly titrated at month twelve is an empirical question with no published answer I could find. Clinicians I would expect to have opinions here have not published them.

Severity migration. A patient who enters at an AHI of 34 — squarely in the range where an appliance is a defensible choice — may sit at 12 a year later on a GLP-1. The device is now treating a different disease state than the one it was prescribed for. Does it get de-prescribed? Retitrated? Nobody has written the protocol.

The discontinuation problem, which cuts the other way. GLP-1 discontinuation rates outside of trials are high, and weight regain after stopping is well documented. A patient who stops the drug and regains the weight returns to their prior AHI, or worse, and now needs a device — possibly one that no longer fits. This is arguably a tailwind for durable, non-pharmacological treatment, and I don't think the device companies have made this argument as forcefully as the data would let them.

Combination therapy. The most likely clinical endpoint, and the least studied. A patient on tirzepatide with residual AHI of 12 is a reasonable candidate for an appliance to close the remaining gap. That is a large market. It is also a market in which the appliance is adjunctive rather than primary, which is a materially different commercial position than the one the decks describe.

I cannot tell you which of these dominates. Neither can anyone else right now, and that is the honest state of the field as of this writing.

What intellectual property actually buys

Vivos's patent estate is not decorative. The company has pursued claims covering the mechanism of its appliance technology rather than merely its physical form, which is the harder and more valuable kind of claim to obtain and the harder kind for a competitor to route around. Paired with FDA clearances that extend to severe OSA — clearances no other oral appliance holds — the regulatory-plus-IP combination is a genuine structural advantage over other appliance makers.

Note the last four words. Over other appliance makers.

A patent is an exclusion right against people making your thing. It is not an exclusion right against people making a different thing that solves the same problem. Kodak's film patents were excellent. The claims were broad, the portfolio was deep, and the enforcement was aggressive. None of that was relevant to what happened, because the threat did not arrive in the form of better film.

This is not a prediction that oral appliance therapy is film. Sleep apnea is not one disease with one cause; a substantial fraction of OSA patients are not obese, and craniofacial anatomy — narrow maxilla, retrognathia, high-arched palate — drives their disease in a way no injectable will touch. That population is real, it is durable, and a device that addresses anatomy rather than weight has a defensible claim on it. That is the strongest version of the Vivos thesis, and it is stronger than the version in the deck, because it doesn't depend on a thirty-year-old prevalence estimate.

What the IP does not do is guarantee the size of that population, and the market-sizing arithmetic — take total OSA prevalence, multiply by the undiagnosed fraction, apply a treatment price — assumes the addressable pool is static. It isn't. GLP-1 drugs are actively subtracting from the weight-driven segment of it while the anatomically-driven segment stays put. Those are different curves, and a portfolio of patents protects a position on the second one considerably better than on the first.

Who this is for, and who it isn't

For investors: the useful question is not whether the patent protection is strong. It is. The useful question is what fraction of the addressable population is anatomically driven rather than weight driven, because that is the fraction the IP actually defends. If you are modeling from an 80%-undiagnosed figure without decomposing it, your model has a 1997 assumption inside it. Ask management for their phenotype breakdown. If they haven't done one, that's the finding.

For clinicians: the actionable gap is retitration. If you are fitting appliances for patients who are on or starting a GLP-1, you are running an uncontrolled experiment on device fit and treatment adequacy, and there is no guideline to fall back on. Repeat sleep testing at six and twelve months on these patients is the conservative move, and somebody should be writing this up.

For patients: if your sleep apnea is anatomical — you're not overweight, you have a narrow palate or a small jaw, you've had it since you were young — weight-loss drugs are unlikely to fix you, and an appliance addressing structure is a reasonable conversation to have with a sleep physician. If your apnea arrived with fifty pounds, the calculus is genuinely different and worth discussing before you spend five figures.

For journalists: the story is not the patent. Patent announcements are routine and companies issue them because they are cheap and positive. The story is that the entire sleep-device sector's market sizing rests on epidemiology from the Clinton administration, and a drug class developed for diabetes is quietly rewriting the denominator.

Who this isn't for: anyone looking for a verdict on whether VVOS is a buy. I don't have a position, I'm not modeling revenue, and the questions I've raised cut in both directions — GLP-1 discontinuation and regain could expand the device market as easily as adoption contracts it.

Back to the number

Eighty percent undiagnosed. It's still on the slide, and every quarter it gets quoted as though it were a standing measurement rather than a photograph.

But it was never a market. It was a finding about how badly a specific diagnostic system, in one American state, was performing in the early 1990s — an artifact of the gap between how many people had a disease and how many doctors were looking for it. It described a failure of detection, and it was converted, through repetition and the ordinary gravity of investor storytelling, into an inventory of waiting customers.

The belief outran its source. And now the population that number described is being altered underneath it, not by better sleep medicine but by a diabetes drug that turned out to make people thinner. Some of those undiagnosed 80% will never present, because their apnea will resolve before anyone looks for it. Others will present in five years, off the drug, heavier than they started, with an airway that no longer matches the device they were fitted for.

A patent portfolio is a good thing to own in either scenario. It just doesn't tell you which one you're in.