DOCTOR EDUCATION · OCCLUSION CONNECTIONS
The implants are succeeding. Everything we attach to them is not.
Studies that follow full-arch cases out to fifteen years still find 95 to 99 percent of them are holding1,2. That part of the treatment is solved. The implants themselves are rarely failing.
What isn’t solved is everything we screw onto them. Depending on which review you read, somewhere between a third and two-thirds of these patients come back1,3 with something broken — chipped porcelain, a loose screw, a cracked framework.
The surgical fixtures integrate and the bone accepts them. But the literature is showing that it’s the restoration on top that is failing.
After thirty years of restoring these cases within our dental profession, the evidence is clear: this is not a materials problem. The evidence says it is not an infection problem either. Dentists are building these cases to a bite position where muscle activity and jaw dynamics are being overlooked and never measured.
The implants are fine. The restoration is not.
Among full-arch prostheses, the single most frequent problem reported is veneer and ceramic chipping — 15–35% of cases, and up to 45.5% in some cohorts.1,2 Screw loosening affects 5–15%1. Framework fracture occurs in under 5% but usually means remaking the entire prosthesis.
Metal-acrylic hybrids fare worst: one thirteen-year analysis found an average of 5.9 complications per patient3 across 87 patients — and 58 of those 87, just under two-thirds, had at least one. Average observation was 42.9 months, with cases followed out to thirteen years. The material choice is measurable on its own: in a systematic review of full-arch cantilevered rehabilitations, metal-resin prostheses had a five-year veneer fracture rate of 37.32%19 against 3.75% for ceramic.
These are not marginal events at the edge of practice. This is the common and typical outcome of implant dentistry done today that a majority of dentists doing full-arch dentistry experience.
So what is missing? Start by ruling out the explanation most of us reach for first.
Sources: Tomar et al., J Prosthet Dent 2026 (umbrella review); Tang et al., Int J Oral Maxillofac Implants 2023; Yilmaz et al., Int J Prosthodont 2023.
The literature does not blame infection. Infection is the easy explanation — the data doesn’t support it.
Dentists are always searching for an excuse for why things are failing. The obvious explanation is peri-implant disease — but that can’t be it. A biologically compromised case that fails does not excuse the more obvious mechanical consequence: breakage. And that falls back on the operator’s understanding and skill in managing the bite.
The data does not support an infectious or biological cause. Cecchinato and colleagues followed 241 implant patients4 and stated the finding plainly: “Technical and biological complications occurred independently of each other.” Technical complications affected 19.5% of patients and marginal bone loss greater than 2 mm affected 14.1% — but only 7.9% had more than one type of complication at all. They were, overwhelmingly, different patients. A prosthesis that fractures, constructed over the implant abutments, is not as a rule a prosthesis that is biologically infected.
It’s a mechanical failure on its own axis. It requires its own explanation.
Source: Cecchinato D, Marino M, Håkansson J, Lindhe J, Derks J. Occurrence of complications in patients restored with implants. Clin Oral Implants Res. 2022;33(9):913–920.
The significance and meaning of odds ratio
The literature predicts mechanical complications in full-arch cases using an odds ratio — how much more likely an outcome becomes when a factor is present. An odds ratio of 5.54 means roughly five and a half times the odds.
Look at what predicts mechanical trouble in full-arch cases and the same names keep appearing:
In the authors’ own words: “Early complications tend to relate to framework design and occlusal overloading, while late complications reflect material fatigue and wear.”
Not every study agrees, and it is worth saying so. A prospective cohort by Coltro found none of these reached significance — not bruxism, not the opposing arch, not the distribution of occlusal contacts or force. The one factor that did was framework design: retention pins under 4 mm. Which points the same way. Their dissenting finding is mechanical too.
There is not one item on that list about bacteria. Every one of them is about load and abnormal forces — how much, from where, and in which direction.
Sources: Tang et al. 2023; Coltro et al., Clin Oral Implants Res 2018; Chen et al., J Dent 2025.
Across systematic review and meta-analysis, implants in patients with probable bruxism fail at roughly twice the odds of non-bruxers — odds ratio 2.189 (95% CI 1.337–3.583)7. A 2025 umbrella review confirmed bruxism as a leading risk factor, with one review reporting an odds ratio as high as 4.688.
And when every risk factor is put into one model, bruxism is the one left standing. In a retrospective evaluation with a mean ten-year follow-up, multivariate analysis found bruxism was the only factor retaining significance18 — not the arch, not the cantilever, not the opposing dentition.
It also shows up in your appointment book. Bruxers in that cohort averaged 7.5 ± 7.9 maintenance visits18 against 3.1 ± 2.9 for non-bruxers (p < .001) — roughly two and a half times the chair time, for the life of the case.
The profession’s response to this is the night guard — Band-aid therapy.
Consider what that means. The strongest behavioural predictor of implant failure is the musculature working at night, and the standard of care is to place a physical barrier between the teeth, indefinitely, and manage the consequence.
Nobody asks the key and underlying question: Why are those muscles hyperactive and working in a destructive manner? Why aren’t they working in a healthy manner?
Sources: Häggman-Henrikson et al., J Oral Rehabil 2024; Barboza et al., J Prosthet Dent 2026;135(2):318.e1–318.e8.
Implants lack periodontal fibers, and the difference is measurable.
A tooth root naturally is surrounded by periodontal ligaments that are densely innervated with Ruffini-like mechanoreceptors11. It is in this root region where the tooth has the ability to sense the load. That ligament detects force, direction, and timing, and reports it fast enough to modify the closing stroke in progress.
An implant is osseointegrated — a functional ankylosis, bone directly against titanium, with no sensing feedback mechanism of the periodontal ligaments. Sensation does not disappear altogether; nerve fibres exist in peri-implant bone and Ruffini terminals have been identified there30. This is osseoperception9,10. But it is not as detailed a feedback system as it is in nature:
| Natural tooth | Implant | |
|---|---|---|
| Minimum detectable force passive tactile sensibility | 11.5 g | 100.6 g31,32 8.75× higher |
| Movement under load shock absorption | 25–50 µm | 3–5 µm |
| Pooled passive thresholds | — | 4–20× higher |
The systematic review states the clinical consequence directly: patients are less able to perceive excessive occlusal forces, which may contribute to overloading risk.
A fair reading of the evidence has to include this. When the patient is actively biting, the gap narrows sharply. In the largest split-mouth trial — 62 patients21,22 — single implants and natural teeth detected foil within 2 to 3 µm of each other, close enough that the authors called them clinically equivalent. Active function recruits muscle spindles, joint receptors and periosteum, and those partly cover for the missing ligament.
But look at what that trial studied: a single implant opposing natural teeth. The compensation it measured depends on a periodontal ligament still being in the system — on the other side of the bite.
And sensation is only one of the losses. The periodontal ligament also drives a protective jaw-opening reflex that pulls the mandible off a damaging contact11. That reflex is absent around an implant. So a biting force the patient does perceive may still not trigger a protective response — thus constant abnormal force exerted on the prosthetic restoration fatigues and eventually breaks.
Put the four together and the picture is different from any one of them alone:
The feedback system that protects a natural dentition from a flawed bite is precisely what the implant patient no longer has.
Why Posterior Occlusal Support Matters — The Neurophysiologic Explanation →
Sources: Hämmerle CH, Wagner D, Brägger U, et al. Threshold of tactile sensitivity perceived with dental endosseous implants and natural teeth. Clin Oral Implants Res. 1995 — the 11.5 g / 100.6 g values, 8.75× higher, p < 0.001. Higaki et al., Clin Oral Implants Res 2014 (meta-analysis, 4–20× range); Song et al., J Oral Rehabil 2022; Kim, Oh, Misch & Wang, Clin Oral Implants Res 2005.
There is one more question in this literature, and it is worth being straight about how settled it is.
Does an implant opposed by another implant feel less than one opposed by a natural tooth? A 2026 systematic review of six clinical studies concluded that it does27. Two others disagree. Urdaneta reviewed the psychophysical literature33 and explicitly rejected the idea that lost proprioception in opposing implants magnifies load, calling it unsupported by the available research; Song’s review did not find opposing dentition9 to be a determinant of osseoperception at all. And no study has ever tested implant-against-implant and implant-against-tooth in the same mouths. The sensory question is unsettled, and this page is not going to pretend otherwise.
The mechanical question is not unsettled.
A natural tooth intrudes 25 to 50 µm under load. An implant moves 3 to 5. In an implant-to-implant pair there is no periodontal ligament on either side of the contact — no give anywhere in the system, and no reflex input from either member. Both halves of the occlusal pair are rigid. Then where does this force distribute its energy to?
That is an All-on-X case. Molar to molar, in both arches.
And the bone data points the same way, whatever the mechanism turns out to be: implants opposed by other implants lost −0.62 mm of crestal bone against −0.20 mm when opposed by natural teeth. Urdaneta, who reported those numbers, attributes the difference to something other than lost proprioception — but the difference is there.
For scale at the other end of the range: complete-denture wearers detect at around 92 µm17, against 10 to 14 µm for implant-supported prostheses. An implant patient is far better off than a denture patient. The question this page is asking is what separates an implant patient from a dentate one.
The case type being sold hardest right now is the one with no shock absorption anywhere in the system. But it is clear based on biophysiology of the masticatory system abnormal force will effect the associated musculature and temporomandibular joints.
Sources: Singh A, Pokarram SS, Podder S. Comparative Analysis of Active Tactile Sensibility in Single Dental Implant and Natural Teeth: A Systematic Review of Clinical Studies. J Prosthet Dent. 2026 (supports the antagonist effect). Urdaneta RA, Leary J, Panetta KM, Chuang SK. The effect of opposing structures, natural teeth vs. implants on crestal bone levels surrounding single-tooth implants. Clin Oral Implants Res. 2014 (bone-level figures; rejects the proprioceptive mechanism). Song et al., J Oral Rehabil 2022 (no antagonist effect found). Mobility figures: Kim, Oh, Misch & Wang, Clin Oral Implants Res 2005. Denture comparison: González-Gil et al., J Clin Med 2022.
The implant literature and research reveal a pattern that dentists must acknowledge. Basic guidelines need to be followed: Narrow the occlusal table.15,16 Shorten or eliminate the cantilever. Use mutually protected occlusion. Reduce contacts in excursion. Add a night guard for bruxers.
These are sensible instructions, and they share a single characteristic: they all manage the accommodated, habitual jaw position — whatever position the case was built to, carrying whatever undiagnosed hyperactive muscle activity came with it. Not one of them asks whether that position is physiologically correct. Not one objectively measures or addresses the biophysiology of where relaxed muscles are supposed to position the jaw. The bite relationships dentists establish for full-arch cases assume the muscles are already in their best relationship — and that assumption is never tested.
You can execute every implant-protection rule flawlessly inside a bite the musculature is fighting. The prosthesis will still express the fight — as chipping, as a loosening screw, as a framework crack and even unresolved muscle tension to the head, neck, shoulders and TMD pain.
Sources: Koyano & Esaki, J Oral Rehabil 2015; Sheridan et al., Implant Dent 2016.
The overload literature is genuinely divided, and it is worth saying so.
Animal studies show that overload on healthy implants does not produce marginal bone loss — in some cases it increased bone-to-implant contact. The only randomised trial found no relationship between maximum bite force and bone loss. The current consensus is that overload is a co-factor that worsens bone loss where inflammation already exists13,14, not an independent cause of it.
That concession costs nothing, because the argument here does not need it. Overload does not dissolve bone. Abnormal muscle activity breaks things — and mechanical failure is exactly what the evidence does support.
But look at how these studies measured the thing they were studying.
Di Fiore and colleagues reviewed seven clinical studies12. Four of the seven found a positive correlation between overload and crestal bone loss. All seven were rated at moderate to serious risk of bias. And — the detail that matters most — only one of the seven used any digital technology to measure occlusal force at all.12
The other six used articulating paper. A wafer of 8 to 200 µm, marking wherever the patient happened to close at that moment, with no reading of how hard, in what sequence, or whether the muscles were relaxed or bracing.
And how good are these instruments, measured against each other? Tested against a silicone criterion standard across fifteen methods, the digital sensor returned a validity of 48 to 56 percent23 — the lowest of the group. Articulating film managed 52 to 63 percent. A 2025 systematic review found digital methods reliable24 for finding where contacts are, but with validity insufficient for contact intensity. A systematic review of implant-supported prostheses concluded outright that paper is not advisable for assessing load intensity on implant crowns25, and that conventional methods are inadequate for precise longitudinal measurement of occlusal change.
And both share the same defect: each inserts a detecting medium between the teeth, altering the very relationship it is recording.
So the studies asking whether occlusal load damages implants were measuring that load with the one instrument that cannot answer the question. That is not an outside criticism — it sits in their own risk-of-bias assessment.
Why Articulating Paper Does Not Reflect Functional Occlusion →
Why Do My Crowns Keep Breaking? →
Sources: Di Fiore et al., J Clin Med 2022; Lee et al., Clin Implant Dent Relat Res 2023; Mojaver et al., J Prosthodont 2025.
There is one more problem, and on full-arch it is the largest of all.
Conventional bite registration sets vertical first — esthetics, phonetics, freeway space — then finds an AP position that satisfies it. GNM reverses that: the relaxed musculature reveals the harmonic AP, and vertical follows mathematically from the proportional relationship — V × 0.618 = AP, trajectory approximately 60° from horizontal — read via the K7 Optimized Scan 4/5 protocol.
And critically:
The two dimensions cannot be determined independently because they are not independent — and they are only two of six.
That lands directly on full-arch. In an All-on-X case the habitual vertical is chosen by the surgical-prosthetic team — space requirements, tooth display, restorative room. Habitual vertical first, by convention.
If AP is mathematically constrained by that choice, then the entire arch is built to an Acquired AP position nobody selected and nobody measured.
For a single crown, getting this backwards is survivable; the ligament absorbs some of the error. On a full arch screwed rigidly from molar to molar, with no ligament anywhere in the system, there is nothing left to absorb it.
And AP and vertical are only the two the profession argues about. The mandible also moves in lateral/frontal, pitch, yaw and roll. A full-arch case gets built to a position in all six, whether or not anyone measured a single one of them, physiologically or not.
So the question is not simply where the jaw closes. It is where it closes in all six. Consider frontal/lateral, pitch, yaw and roll of the mandible — pathologic or physiologic position and relationship.
The AP-First Sequencing Principle in GNM Bite Registration →
Scan 4/5 with resting EMG — the position read, not chosen.
Scan 4/5 and the resting EMG above are two of a much larger set. The K7 records the mandible through every function that matters — how it travels, how it chews, how it clenches, and where the teeth first meet. You do not need to read all of these today. What matters is that each one is a measurement rather than an assumption.
Scan 2 — jaw tracking, terminal contact
Scan 8 — functional chew cycles, before and after GNM
Scan 11 — functional clench EMG
Scan 12 — first tooth contact EMG, before and after
A natural tooth reports an occlusal interference at roughly 20 µm. An implant needs about 48 µm before the patient knows anything is wrong. Surface EMG identifies an occlusal contact at the 10–20 micron level, sensitive to 5.5 µV during a functional clench.
The instrument detects what neither the implant nor the patient can report.
On a natural dentition, objective measurement is a refinement — the ligament is still there, still reporting, still protecting and still accommodating. On an implant case it is something else entirely: these diagnostic measurements enhance the doctor’s awareness and improve the odds beyond the subjective “how does it feel?” Measuring and recording the patient’s bio-physiologic jaw position and functional muscle EMGs at a higher level of proprioceptive occlusal detail during “tap tap tap” or during the functional chew allows the clinician to see the load levels that an implant patient is no longer able to sense.
And the load does not stay where you left it. Using digital occlusal analysis, relative occlusal force on implant crowns rises significantly from two weeks to three months, keeps climbing through twelve to twenty-four months28,29, and by five years has exceeded the natural control teeth — the authors of that study raise overloading as the concern. The force on the implant goes up while the patient’s ability to feel it stays diminished. Those two lines move apart for years after you hand the case over, which is why the reviewers themselves recommend routine occlusal monitoring at follow-up.
Worth being straight about where that finding comes from: those are digital-sensor studies — the best instrument the profession currently has for tracking occlusal change over time, and one whose own validity for locating contacts measures 48 to 56 percent. Its findings point at a problem it cannot itself fully characterise.
That is the argument for measuring before you restore and again after delivery of the implant prosthetics. It is arithmetic, not philosophy.
At Occlusion Connections these measurements are taken with the Myotronics K7, and the scans are read complete, top to bottom — not sampled for the one number that supports a conclusion already reached.
The K7 readout: jaw tracking, EMG and ESG on one record.
Consider how provisionals are normally used.
The dentist prepares or removes the teeth first. Then the prepared teeth are temporised — and those temporaries are called provisionals. In the clinician’s mind the patient has now been provisionally placed, along with the bite, into some assumed position, in the hope that the patient will settle down and stop complaining. When the complaints quieten, that assumed position becomes the final.
That is all backward.
The irreversible step came first. The position was never determined — it was inherited from whatever the preparation left behind, and then ratified by the patient’s eventual silence.
The physiologic position has to be found first, as the diagnostic blueprint. From that stable, measured position, you then work out how to restore or reposition the teeth to it. The order is not a preference; it is the difference between building to a known position and hoping your way into one.
A minimum of three months stable, pain free, and off medications — and some cases take considerably longer, depending on the condition the patient arrived in.
Stability is not the patient saying they feel better. It is comfort plus objective measured recordings confirming the position is holding: the musculature quiet, the closing path repeatable, the scans reproducing.
Once the patient is comfortable and proven stable on those recordings, you can proceed definitively — restoring to the proven position and the proven dimensions, with assuredness.
Not the other way around.
With GNM there is no longer any need to be guessing the bite.
On a full-arch case this matters more than anywhere else. A milled, screw-retained prosthesis is expensive, rigid, and difficult to revise. It should be the last thing that happens to a position — not the instrument used to discover it.
Implant dentists are better positioned to act on this than anyone in the profession. You already plan digitally, verify with imaging at every stage, and think in terms of load, vectors, and tolerances. Measured occlusion is that same discipline applied to the one thing most implant workflows still decide by habit — where the jaw closes.
The field is already moving. A 2026 study designed maxillary full-arch implant rehabilitations26 using individualised jaw motion data rather than a conventional static digital workflow, and measured the difference in adjustment depth, adjustment area and chairside time. Jaw motion, integrated into full-arch prosthetic design. That is the direction this article has been describing.
The surgical side of implant dentistry became predictable when it became measured. The occlusal side has not yet made that transition.
Read next
Train with us
Occlusion Connections teaches the measurement protocol — TENS, jaw tracking, EMG, ESG — and its application to restorative and full-arch sequencing.
The entry point is Level 1, the foundational course where the measurement discipline is built. Levels 2 through 4B extend it.
COURSES ARE AGD PACE APPROVED
Updated: August 5, 2026
Written by Clayton A. Chan, D.D.S. — Founder and Director, Occlusion Connections | Las Vegas, Nevada
Occlusion Connections — The Center for Gneuromuscular Dentistry and Orthopedic Advancement
6170 W. Desert Inn Rd., Las Vegas, NV 89146