Written by Ryan Nord, MD — Sleep Surgeon, VCU Health

Data collected February 27 – August 1, 2026 · complete closed dataset — collected under Decision Tool V1.0, retired August 2, 2026

What mattered most, in their own ratings

Each person moved a 0–10 slider for each of nine factors. Below is the average rating each factor received from the people who rated it — the most direct answer this dataset gives to “what do patients actually care about?”

Bars run the full 0–10 scale — the gaps are as small as they look. Between 131 and 144 people rated each factor; the counts differ because not everyone moved every slider. The bracketed groups are statistical ties — the gaps inside them are smaller than the wobble produced by any reasonable way of cleaning the data, so their internal order means nothing.

The first thing to say about this chart is how flat it is. Every one of the nine factors averaged between 6.8 and 8.0 out of 10 — a spread of about one and a fifth points across the whole list. People weighing this decision do not want to trade anything away. They want the experienced surgeon and the MRI access and no battery surgeries, and they say so at nearly the same volume for almost everything you put in front of them. That is worth knowing before you walk into a consultation expecting a patient to have a single overriding concern.

One factor does separate itself. Surgeon experience is the only item with clear daylight around it — 70% of the people who rated it gave it an 8 or higher, and 60 gave it a flat 10. As the surgeon on the other side of these conversations, that is the part I keep coming back to: the single highest-rated factor is not a device specification at all. Below it, the practical live-with-it-for-decades concerns — MRI access, incisions, battery replacement — sit together in a tie, and the lowest-rated item on the whole list is the preference for a fully internal device with nothing worn at night, at 6.8.

What tipped each person’s result

Separately from the raw ratings, the tool recorded the five factors that contributed most to each person’s recommendation. That is a different quantity: a factor lands here when a high rating meets a feature where the two devices differ, so it reflects the tool’s design as much as the person’s answers. Read it as “what separated the two devices for this person,” not as “what they cared about.”

Counted out of all 151 respondents. Each person contributed up to five factors; the 14 whose result came back a tie contributed none.

One number here is an artefact, and it should not stand without correction. Back-sleeping fit placed in only 4 people’s top five — but the same people rated it 7.2 out of 10, statistically tied with the evidence record and device bulk. The reason it almost never placed is that V1.0 gave it the smallest weight on the list, so even a 10 out of 10 on back-sleeping counted for less than a middling rating on surgeon experience. It could not have ranked highly here no matter what anyone said. That is a fair description of the clinical reality — the two devices differ less on positional response than on MRI access or battery replacement — but it is a statement about the instrument, not a finding about patients. Anywhere else on this site where you see back-sleeping described as something patients don’t care about, this dataset says otherwise.

Preferences split almost evenly between the two implants

The tool weighs each person’s stated priorities — not brand — and neither device dominated. That’s notable given Genio only received FDA approval in 2025:

“Leaned” combines strong, moderate, and weak results for each device: 48% Inspire, 42% Genio, 9% no clear lean.

A caveat that belongs with that split rather than buried in the methods. V1.0 was too willing to call a result decisive: 110 of the 151 responses — 73% of everyone, and 80% of everyone who got any lean at all — came back with its strongest possible verdict. A single emphatic slider could push it there. The direction of the split is trustworthy; the confidence V1.0 expressed about each individual result was not, and rebuilding the tool so that no one answer can dominate was the main reason it was replaced.

Who answered

The demographic questions are optional, so each breakdown below reflects only the people who answered it (the count is shown with each chart). The picture they paint: exactly the audience hypoglossal nerve stimulation exists for.

Among those who answered the remaining questions: 67 male and 43 female of the 110 who gave a sex at birth; insurance was mostly commercial (48 of 74), with Medicare (14) and Medicaid (6) next; and of the 87 who said how they found the tool, a Google search (23) and a friend (22) were neck and neck, with their own doctor third (18). Responses came from 27 US states — Florida (10), Texas (9) and Virginia (7) most often — plus three from the UK and EU. Thirty people (20%) reported a skin-adhesive sensitivity, 29 mild and one severe, which is a practical constraint on any device worn against the skin at night.

One more pattern worth recording, because it is about the site rather than the patients: this dataset is heavily weighted toward its final weeks. July alone produced 85 of the 151 responses (56%), against 9 in February and 5 in May, with the single busiest day — 15 responses — on July 15. The audience arriving at these pages grew faster than the collection window, so the later responses carry more of this data than the earlier ones do.

Methods & limitations

• What this is: responses collected, without names or contact details, through the decision tool on this site between February 27 and August 1, 2026. No names, emails, or contact information were collected, and demographics are broad ranges only, reported here in aggregate. A response was logged only if the consent box was selected at submission and at least one priority was rated. In V1.0 the box came pre-selected, and the tool required it to generate a final recommendation — so this dataset reflects everyone who completed the tool, rather than a subset who opted in separately. The rebuilt tool used the same pre-selected box until September 3, 2026; since then it shows the result either way and stores answers only if the person checks a box that is unchecked by default. What is stored, and how to have it deleted, is on the privacy page.
• Cleaning: from 236 raw submissions I excluded 21 duplicate double-clicks, 32 responses from a single unexplained same-day traffic burst on March 19, and 3 of my own build-day test entries. That left 180. I then found 15 sequences where one person re-ran the tool several times in a single sitting — submissions seconds apart with one or two sliders nudged, and in one case four submissions with identical ratings while the person filled in the optional demographics. Counting those as separate people would let one visitor’s opinion count four times, so only the final submission of each sequence is kept, removing 29 more rows. The 151 reported here are 151 different people.
• How the averages are calculated: each slider started at 0 and had to be moved, so a zero cannot be distinguished from a slider the person never touched. Averages are therefore taken over the people who rated that factor above zero, and the number of raters is given with each figure. As a check, recomputing on only the 96 people who moved all nine sliders leaves the bands below intact.
• Why bands instead of a ranking: the middle of the list is a statistical tie. If I keep the first submission of each repeat sequence instead of the last, or leave the repeats in entirely, the top two hold and the bottom two hold, but the factors inside each bracketed band swap places. Their order is not a finding, so the chart doesn’t draw one. Leaving all 29 repeats in gives 180 responses, a 49% / 41% / 10% device split rather than 48% / 42% / 9%, and moves no average by more than about a quarter of a point.
• Two different measures: the ratings chart reports what people said. The “what tipped each result” chart reports the tool’s own top contributors, which combine a person’s rating with a fixed weight for how far apart the two devices sit on that feature. Those weights were mine, and they are why back-sleeping fit — rated 7.2 out of 10 — almost never appears there. Do not read the second chart as a ranking of patient priorities.
• Collection is closed. V1.0 was retired on August 2, 2026 and replaced by an instrument that asks different questions and scores them differently. Its responses are stored separately and the two sets will never be pooled, so these 151 are final rather than a snapshot that will grow.
• What this is not: a scientific study. It’s a convenience sample of people who found one educational website and chose to complete a questionnaire. It skews toward motivated online researchers, toward the weeks when the site’s traffic was highest, and the optional demographics reflect only those who answered. Treat it as counseling insight, not evidence.
• Disclosure: I consult for Inspire Medical Systems, and I have been a consultant for or received research support from Nyxoah (Genio’s manufacturer), LivaNova, and Avivomed — full disclosures are on the About page. The tool scores each person’s stated priorities against published device characteristics — the scoring logic never favors a brand, and this page reports what visitors said, unedited.