What is the Inspire success rate?
Published success rates for this implant run from under half of patients to over ninety percent. Most of that spread is not the device. It is how the follow-up sleep study was done, and where the finish line was drawn.
On this page
The short answer. The Inspire success rate most worth quoting comes from the trial that led to FDA approval: 66% of patients — 83 of 126 — met the definition of success at one year, meaning their apnea events were cut by at least half and brought under 20 an hour.[1]
That is the number I start with, because it comes from the most rigorously conducted study we have. But it is one number produced by one method, and the full picture is wider than that. Published success rates for this device run from about 46% to 91%, and most of that range has nothing to do with how well the implant works. It reflects two choices made after surgery: what kind of sleep study measured the result, and how low the apnea number had to fall to count.
This page explains both, and tells you which number I think you should actually judge this operation by.
Success is a definition, not a fact
When a surgeon tells you an operation works 70% of the time, that figure is doing a lot of quiet work. Somebody had to decide what counts as working, when to measure it, and how. Change any one of those and the number moves — sometimes by more than the difference between two entirely different operations.
For sleep apnea surgery the convention comes from a 1996 analysis by Sher and colleagues, and it has two parts. To be counted a success you must (1) cut your apnea–hypopnea index — your AHI, the number of times an hour your breathing stops or shallows — by at least half, and (2) end up below a set threshold.[2] Both parts have to be true. Halving a very severe AHI from 80 to 40 is not a success under this rule, and neither is drifting from 22 down to 19.
That is a sensible definition. The trouble is that different studies pick different thresholds, and different clinics measure the AHI in ways that are not comparable.
Why the published numbers disagree
Here is what good studies actually report. These are not cherry-picked outliers — they are the pivotal trial, the largest registry, the biggest meta-analysis, and the studies that measured the same patients two different ways.
| Study | Patients | How the result was measured | Success rate |
|---|---|---|---|
| STAR trial, 12 months[1] | 126 | In-lab sleep study | 66% |
| STAR trial, 5 years[3] | 71 of the original 126 | In-lab sleep study, volunteers only | 75% |
| STAR trial, 5 years, counting everyone[3] | 126 | Same, but non-attenders counted as failures | 63% |
| Pooled analysis of 39 groups[4] | 3,220 | Mixed — whatever each center used | 74% (range 27–96%) |
| Same patients, two ways[5] | 61 | Titration study → home study | 80% → 46% |
| Same patients, two ways[6] | 148 | Best setting → whole night | 87.8% → 68.9% |
| Same patients, two ways[7] | 43 | Titration study → home study | 91% → 52% |
| My own patients[8] | 111 | Mostly all-night home studies | 77.5% |
Note the last three rows. Those are not different groups of patients being compared — they are the same patients, measured twice. The device did not change between the two columns. Only the sleep study did.
The pooled analysis is worth pausing on. It gathered 39 separate patient groups and calculated an average success rate of 74%. But it also reported the range you should expect the next study to land in, and that range ran from 27% to 96%.[4] When the spread is that wide, the average is not telling you much.
Five of the six authors of that pooled analysis have financial ties to one or both implant manufacturers, and the senior author is an employee of one of them. I still cite it, because it is the most complete synthesis available and its headline finding — that the studies disagree enormously — runs against commercial interest.The titration study problem
This is the single most important thing on this page, and it is rarely explained to patients.
After your implant is turned on, you need a sleep study to see whether it is working and to find your setting. There are two ways to do it, and they do not produce comparable numbers.
A titration study is done overnight in a sleep lab. A technician gradually raises the stimulation, watching how your breathing responds, hunting for the best setting. The number that usually gets reported from that night is the lowest AHI achieved at the best setting the technician found.
An efficacy study — often an all-night home sleep test — leaves the device on one fixed setting and records the whole night: every position, every sleep stage, including the REM sleep late in the night when apnea is usually at its worst.
One measures your best moment. The other measures your night.
A 2025 study from Thomas Jefferson quantified the gap precisely. Sixty-one patients had both tests. Measured on the titration study at the ideal setting, the average AHI was 8.8. Measured on an all-night home study, the same patients averaged 17.6 — twice as high. The success rate fell from 80% to 46%.[5]
The explanation is almost mechanical. During that titration night, patients spent an average of just 50 minutes asleep at the ideal setting — about 15% of their total sleep.[5] The reported number came from that fragment. The home study covered nearly seven hours. And when the researchers compared the two tests at the same device setting, the difference vanished.[5] The device was performing identically. Only the measurement differed.
Two other groups found the same thing. In 148 patients, scoring at the optimal setting gave 87.8% success; scoring the entire night in those same patients gave 68.9%.[6] In an earlier series of 43, the split was 91% versus 52%.[7]
This matters for reading the STAR trial too. STAR used full-night studies, which is one reason its 66% is more conservative than many single-center reports you will find quoted.[1][5] When a practice advertises a success rate well above STAR’s, the first question is not whether their surgeons are better. It is which sleep study produced the number.
What I found in my own patients
I published my own results, because I think you are entitled to see them rather than only the manufacturer’s.
Across 111 consecutive patients I implanted, the average AHI fell from 32.8 to 10.4, and 77.5% met the success definition.[8] Average device use was 6.4 hours a night.
Two things about that number deserve stating plainly. First, I used the stricter finish line — under 15, not under 20 — which I explain below. Second, I measured with all-night home sleep tests wherever I had them, falling back on titration values where I did not. In the paper itself I named that mixture as a limitation, and we are now collecting all-night home studies prospectively so that future results rest on one method.[8] I would rather tell you that than quote you a cleaner-looking figure I could not stand behind.
That study also found something worth knowing if you have been told your airway collapses from side to side: patients with complete lateral wall collapse succeeded 66.7% of the time, against 84.8% for everyone else.[8] It is a real disadvantage, and it is not a disqualification.Where the finish line is drawn: Sher 20 and Sher 15
The second reason published numbers disagree is that studies do not agree on how low is low enough.
The original Sher definition requires the final AHI to be under 20.[2] That is what STAR used, and it is what most published trials use. You will see it written as Sher20.
A growing number of us use under 15 instead — Sher15. Same 50% reduction requirement, lower ceiling. It is a harder test, and it produces lower success rates on identical patients. In the Jefferson study, moving from the under-20 bar to the under-15 bar dropped the measured success rate by a few points at every measurement method.[5]
Why does a few points of ceiling matter? Because of where those thresholds sit on the clinical map:
- Under 5 — considered normal.
- 5 to 15 — mild sleep apnea.
- 15 to 30 — moderate sleep apnea.
- 30 and above — severe sleep apnea.
A patient who ends at an AHI of 19 is a success under the under-20 rule. But 19 still sits inside the moderate range. They came in with a disease and they leave with a milder version of the same disease, counted as a win.
The under-15 rule requires you to land in the mild band. That is a meaningfully different promise.
Why I use the stricter bar
Getting patients into the mild range is, to my mind, the whole point of sleep surgery. The reason comes from a study that has nothing to do with implants.
The Wisconsin Sleep Cohort followed 1,522 ordinary adults — not patients, not people referred to a sleep clinic — for up to eighteen years, and recorded who died.[9] It was not funded by industry. It remains one of the cleanest looks we have at what untreated sleep apnea does to a person over a lifetime.
Adjusting for age, sex and weight, people with severe apnea were about three times as likely to die during follow-up as people with none.[9] Among those who never used CPAP, that rose to 3.8 times.[9] The gap did not depend on feeling sleepy: people with severe apnea who reported no daytime sleepiness carried the same elevated risk as those who did.[9]
Now look at the left three bars. Mild and moderate apnea were not statistically distinguishable from having no apnea at all in this study.[9] The clear, unmistakable signal sits in the severe group.
So the first job of this operation is to get you out of severe. Both the under-20 and the under-15 definitions do that. But if I am choosing where to set the finish line, I would rather aim for the band where the risk curve is flat than stop at the edge of moderate disease. Under 15 puts you in the mild range. Under 20 can leave you at 19.
Being straight about the limits of that argument: Wisconsin does not show that an AHI of 12 is safer than an AHI of 18. Mild and moderate were statistically similar to each other, and to no apnea at all. My preference for the stricter bar is a clinical judgment about aiming for the safer end of the map, supported by the direction of this data — not something the study proves. And this is untreated apnea observed over time. No study has yet shown that lowering your AHI with any treatment, including this implant, lowers your risk of dying.What to ask your surgeon
If you take one thing from this page, take these three questions. They will tell you more about a quoted success rate than the number itself.
- Which sleep study produced that number? A titration study at the best setting, or an all-night study on my actual setting?
- Where was the finish line? Under 20, or under 15?
- What was the denominator? Everyone implanted, or only the patients who came back for testing?
When you ask me those questions, the answers are: mostly all-night home studies, under 15, and everyone with complete data in the study period — 111 patients, 77.5%.[8]
And whatever the number, remember what it is measuring. It is a threshold on a sleep study. It is not the same as how you feel, how your partner sleeps, or whether you would do it again — all of which matter, and all of which I would rather discuss with you in clinic than reduce to a percentage.