OMFS Implants Study Day · 7 August 2026 · 09:00 Foundations
Foundation of implants
Rawand Shado BDS AFHEA
Nine complications, walked backwards to the eight principles
underneath them.
Fifteen seconds. It is a title. Do not start teaching off it — the next slide sets the method and the licence position, and the first photograph is the one after that. If the poll server is running, this is the moment to check the address on the next slide's code resolves from your own phone before nineteen other people try it.
how far back
09:00 · Foundations · nine walk-backs ·
shortest walk first
We start at the disaster and reason backwards.
1Osseointegrationnot yet
2Primary stabilitynot yet
3Bone and graft healingnot yet
4The soft tissue sealnot yet
5Load transfer without a PDLnot yet
6Three-dimensional positionnot yet
7Biofilm and the peri-implant lesionnot yet
8Anatomy and its limitsnot yet
Every case starts at the complication and walks
backwards — one decision at a time, to the root cause. You commit to
a cause before you are told anything.
The question is never who got this wrong. It is at which point the
outcome stopped being avoidable.
Observed what the published record actually states. Inferred everything else — by their authors too, who wrote them backwards as well.
Scan once — it follows the deck
Every case report here is open access — published under Creative
Commons, and we hold the right to show the images. Each carries its credit
on the slide.
AK1.3
+0:00. Ninety seconds, and not a second more — the room has not seen a photograph yet. Get every phone on the code here; if you chase scans later you will do it during the reveal and lose the silence. Point at the bar along the top (how far back) and the eight dark cards (where we have to get to). SAY ALOUD, because it is no longer on the slide: eight foundations, and by 10:00 every one of them is lit, none of them by a lecture — the nine are ordered by how far back you have to walk, this one is thirty seconds and the last is before she sat down — voting is anonymous and compulsory, and being wrong is the mechanism of this hour rather than an embarrassment in it. Also say the two standing declarations here: this day claims Miller level 3 at most, 'shows how', and no ILO claims level 4; and industry supplies the kit and none of the content. Do NOT preview the case.
how far back
today · thirty seconds ago
Walk-back 1 of 9 · UL7
· a composite patient — two published cases
She is 75. This was the sixth and last implant of the morning.
A · Kasapoglu 2026“Periapical radiograph showing the dental implant displaced
into the left maxillary sinus” — the paper’s own caption.
Kasapoglu et al. 2026. Reproduced under CC BY 4.0.
Full arch, six implants, freehand, one visit. The last one turns,
and then it turns against nothing.
Observed a periapical taken minutes after the event. Inferred nothing yet — not the bone height, not the brand, not the length.
Six implants, one visit, freehand. Nothing about this morning was
unusual until the sixth one turned.
+1:30. Put the film up in silence and count to five before you say anything. The room will laugh — let it, that recognition is why this one opens the hour. Give the history in the order written and then STOP. Every extra fact you add makes the vote on the next slide worth less. NOT ON THE SLIDE, and worth one sentence if the room asks: in the published case the receiving unit could not obtain the implant's brand, diameter or length from the referring clinic at all.
how far back
today · nothing revealed yet
Walk-back 1 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe operator pushed too hard on the last turn—
BThe site never had the bone to give it grip — it was gone before it was seated—
CThe osteotomy was overprepared and the implant had nothing to bite—
DThe sinus floor was thin and it was perforated on the way in—
Hold on to what you picked. Two of those four are the same answer one rung
apart — and one of them is where this walk-back ends.
+3:00. THE STEENHOF EVIDENCE IS NO LONGER ON THIS SLIDE and it does not need to be — the closing slide carries it, once, as the hour's own claim about itself. If you want it here in a sentence: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm. Chase abstainers — commitment is the whole mechanism, and this is the first vote of the day so the habit is set here. The bars are hidden until you advance one step, so nobody anchors. Read the distribution out and say nothing about it. Poll id w1. If the phones cannot reach the laptop (NHS guest wifi very often isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect a heavy vote for A. It is the M&M answer, it is unfalsifiable, and it is the one the rest of the hour is designed to walk past.
how far back
thirty seconds earlier · the driver is still on it
Step back one · thirty seconds earlier ·
the implant is being seated
Nothing pushed it in. There was nothing holding it out.
B · Bär 2024Seatedgood stability recordedB · Bär 2024Four monthsin the antrumB · Bär 2024Seventy-two hoursin the ascending colon
Bär’s Cases 1 and 3, cropped — different published
cases from the one being narrated. Bär et al. 2024. Reproduced under CC BY 4.0.
Type IV bone under a pneumatised antrum. The threads had nothing
to cut into.
Observed the implant is in the antrum, and it went there while it was being seated. Inferred that it had no grip — no torque and no ISQ was ever recorded.
321 published cases, and not one denominator. If anyone gives you a
percentage for this, ask what was on the bottom of the fraction.
+5:30. This is the rung the walk-back exists to reach, and it arrives early because the walk is one step. Let the three panels run before you say anything — seated, in the antrum, in the colon — the room does not need the middle explained. Land the denominator line hard — it sets the evidential standard for the next eight walk-backs. THE FULL EVIDENCE, if challenged: lack of primary stability heads Bär's published risk-factor list, ahead of peri-implantitis and intrasinus pressure; Kasapoglu names sinus pneumatisation, reduced residual bone height and poor bone quality; Bär and Seigneur both record that actual incidences are unknown. And 87% of displacements are found AFTER the operation, not during it — of those with a timing recorded, 63% went within six months.
how far back
thirty seconds earlier · the driver is still on it
4 roads out of this junction. 1 taken.
How you would have known, before you let go of the driver
Taken — 1 of 4
Tactile judgement at the driver
Available, not taken
Insertion torque, in NcmResonance frequency analysis — ISQPeriotest
Twenty seconds. Point at the lit chip FIRST — tactile judgement at the driver is the road that was taken, and it is the only one of the four that leaves no number behind. Then the three that were not: the instruments in the unlit group measure different physical quantities, which is why they disagree with each other. Tactile judgement is not a bad measure. It is an unrecordable one, and that is the whole of this junction. Do not read the group headings out.
how far back
before the drill · the day it was planned
Step back two · before the drill ·
we have just walked past a rung without stopping — the osteotomy
The height was measured. It was measured on a panoramic.
B · Bär 2024The planning panoramic. A different published case —
Bär’s Case 1, 5 mm of residual bone, a 5.0 × 8 mm implant, good
primary stability recorded, uneventful healing. Bär et al. 2024. Reproduced under CC BY 4.0.
A panoramic is a projection. It gives you a height and nothing
about width, density, or a septum standing in the middle of the site.
Observed the height was measured — 5 mm, on a panoramic. Inferred that no cross-sectional imaging of this site ever existed.
A number taken in the wrong plane is not a rougher version of the right
measurement. It is a different measurement.
+9:00. Say why we walked past the osteotomy without stopping: undersizing and osseodensification buy grip in bone that is there, and this site did not have bone to undersize. The line to leave them with is the crimson one. NOT ON THE SLIDE: in the same series a planning CBCT showed less than 1.5 mm at UL7 with poor initial stability noted at insertion, and that one went too; placement below 4 mm residual height is on the published risk-factor list.
how far back
before the drill · the day it was planned
18 roads out of this junction. 3 taken.
How this site was looked at before the drill went in
Taken — 3 of 18
Clinical examination and study castsPanoramic radiographFreehand placement
Available, not taken
Diagnostic wax-upPeriapical radiographCBCT — field of view and doseRadiographic stent / dual scanDICOM and STL alignmentProsthetically-driven planningSafety margin — maxillary sinusSafety margin — nasopalatine canalInterimplant and implant-to-tooth distancePilot-drill guidePartially guidedFully guided staticDynamic navigation
Thirty seconds. Point at the three lit chips first — an examination, a panoramic, and a freehand drill — then at the count, and let the room see how much of the middle group was available on the day and was not used. The two dashed chips (robotic, AI-assisted) are there because they exist, not because the evidence puts them here; say that if anyone asks, and do not read the list.
how far back
before anything was raised · the site as it was
Step back three · before anyone picked up an instrument
· the floor of this walk-back
Type IV bone under a pneumatised sinus.
It was never going to hold it.
Short implants match sinus lift on survival, and beat it on bone
loss. The graft is not the safe default.
Observed a posterior maxilla, and three series naming the same setting. Inferred that this site could not have held this implant — nobody randomised this ridge.
The further back you have to walk, the less it is about the drill and the
more it is about the patient — and this was the shortest walk of the
nine.
+12:00. This is the one rung of the hour with no photograph, and it is the floor of the walk-back — land the crimson half of the headline and let the room sit in it for a beat before you advance to the roads. THE NUMBERS, if you want them: Bär describes the atrophied maxilla as progressive pneumatisation, thin residual bone and typically type IV density on Lekholm & Zarb; septa prevalence runs 20-58% across cohorts, so the sinus floor is not reliably flat and a panoramic will not tell you; Mester 2023 and Alenezi 2025 both find short implants non-inferior on survival and better on marginal bone loss and biological complications.
how far back
before anything was raised · the site as it was
17 roads out of this junction. 1 taken.
What you could do with a posterior maxilla that will not hold a standard implant
Forty seconds — the longest junction of the three, and the only slide in this walk-back the room is allowed to read rather than glance at. Point at the FIRST chip in the unlit group before you point at the lit one: placing nothing is a road, it is on the map, and it is almost never written in the notes. Ask by show of hands who records the non-implant option in their consent for a posterior maxilla. Then move — do not let this turn into a short-implant advert.
how far back
the site as it was · and the vote you cast before you saw it
The ballot, answered · walk-back 1 of 9
You voted before you were told anything. The walk ended at B.
AThe operator pushed too hard on the last turn
BThe site never had the bone to give it grip — it was gone before it was seatedwhere the walk ended
CThe osteotomy was overprepared and the implant had nothing to bite
DThe sinus floor was thin and it was perforated on the way in
Nothing pushed it in — there was nothing holding it out. Type IV bone under a pneumatised antrum gave the threads nothing to cut into, and no torque and no ISQ was ever recorded, so “too much force” is an answer nobody can check.
Observed type IV bone under a pneumatised antrum, and no torque or ISQ ever recorded. Inferred that the operator pushed too hard on the last turn.
Twenty seconds, and the first time in the hour anyone is told they were wrong. If the poll server has been running, go back one press on the ballot slide or just read the room's split off the bars you already showed — say the distribution out loud before you name the option. Then name it: the site never had the bone to give it grip; it was gone before it was seated. Most rooms put their weight on the operator pushing too hard, which is the M&M answer — unfalsifiable, and the one the rest of the hour is designed to walk past. Do not defend the walk and do not take questions here; the next slide is the principle and it answers them. Move.
how far back
the site, before any of it · day zero
The floor · foundation 2 of 8 ·
primary stability
Mechanical grip buys the time biology needs.
Present at the moment of seating and at no other time. It holds
the implant still through the trough, while biology rises to replace it.
Observed grip is necessary. Inferred that it is sufficient — Bär’s Case 1 had 5 mm, adequate stability, and went anyway.
Two of eight lit, and the walk was one step long. Eight walk-backs to go.
+14:30. Let the headline sit before you speak, then advance once and let the board fill under it. If you say one extra sentence, say this: every loading protocol in implant dentistry is downstream of this one number, which is why it is foundation 2 and not a footnote to foundation 1. The amber line is not a hedge — a registrar who leaves thinking torque predicts survival has taken the wrong thing; in Bär's Case 1 the height was 5 mm, stability was adequate, healing was uneventful, and the implant was in the sinus four months later, and the authors write that the location seems to be the risk factor. ASK THE ROOM to predict which card lights next — that prediction is the learning. WHAT CHANGES ON FRIDAY, say it, it is not on the slide: write the residual height in the notes AND the plane you measured it in; record what the implant actually gripped, and if you cannot get a number that is itself the finding; and if one does go in, image it in three dimensions before you go looking for it, not after. Then hand over: ten minutes, and her tongue is against her palate.
how far back
Definitions · foundation 2 of eight ·
it has been earned; now name it
Primary stability
Primary stabilityThe mechanical grip an implant has at insertion, before biological healing has created new bone.
Insertion torqueThe rotational resistance measured while the implant is seated, recorded in Ncm.
ISQA unitless resonance-frequency measure of the stiffness of the implant–bone complex.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
Definitions · foundation 8 of eight ·
it has been earned; now name it
Anatomy and its limits
OPG · panoramic radiographA two-dimensional projection of both jaws. It shows height, but collapses depth.
CBCTA cone-beam X-ray volume that shows the site in three dimensions.
Residual ridgeThe alveolar bone remaining after teeth have been lost and the socket has remodelled.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · ten minutes after the pilot drill
Walk-back 2 of 9 · LR6
· a composite patient — two published cases
Ten minutes ago you were drilling. Her tongue is against her palate.
She is 58, fit, no anticoagulant. 13 mm above the canal
with a 10 mm implant.
+3 min — brisk bleeding that pressure will not stop
+6 min — the floor of her mouth is firm, and rising
+8 min — she has stopped swallowing her own saliva
+10 min — her tongue is against her palate
There is no photograph on this slide, and that is a finding
rather than a gap. The only published image of this appearance in this corpus
carries a non-commercial, share-alike licence.
This deck reproduces CC BY or nothing, so you get the description and the clock
now, and the cross-section two slides from here.
Observed bleeding, a firm and rising floor of mouth, a tongue against the palate. Inferred nothing — not the vessel, not the shape of this jaw.
This is an airway. Nobody here drilled it. Everybody could
have.
+0:00. Read the four lines of the clock out, in order, and then stop. Do not say the words haematoma, lingual or artery — the room has to arrive at those itself, two slides from here. If someone calls it out, thank them and move on without confirming. NOT ON THE SLIDE, and worth saying as you read the clock, because it is the texture that makes it real: at six minutes she is uncomfortable rather than frightened and she can still speak; at eight the swelling is across the midline and the tongue is lifting off the floor; at ten she is sitting forward, she will not lie back, and her voice has changed. The history matters because of what it RULES OUT — she is fit, she is on nothing, and one of the four ballot options is about to say she was anticoagulated. The absence of a photograph is worth ten seconds: the one that exists is licensed non-commercially and share-alike, and we do not reproduce figures we cannot re-use or crop. That is the same rule that decides what goes on every other slide today, and it is the honest version of 'we could not find a picture'.
how far back
today · the emergency, not the cause
Still today · the emergency, before the cause
The first move is not a diagnosis. It is a phone call.
Ten minutes in, the diagnosis can wait. The airway cannot.
Law 201725 published cases · emergency intubation or tracheostomy68%
Review of 18 casescited by Hwang 2013 · intubation or tracheostomy used during haemorrhage control89%
Law 201725 published cases · required surgical management in hospital72%
Two independent reviews of the published cases, six years
apart. Law found 25 reported cases in twenty-six years.
The floor of the mouth is lax, so blood spreads instead of collecting.
There is usually nothing to drain — and Barrientos-Lezcano reports that
opening it can destroy the self-tamponade and restart the bleeding.
Observed of the cases written up, most needed an emergency airway. Inferred that this is the risk per event — a case series has no denominator.
+1:15. THE ONE EXTRA SLIDE IN THIS WALK-BACK, and this is why: it is the only slide in the whole hour where the right answer is a phone number rather than a thought. Say that out loud. Resist the urge to run an airway tutorial — the teaching point is the ORDER, not the technique. The one line that must be said is the crimson one: opening a lax floor of mouth can destroy the tamponade and restart the bleeding, so 'do something' is not automatically safer than 'get help and protect the airway'. NOT ON THE SLIDE, and this is the fan that came off it: there are 14 roads out of this junction and only 2 of them are yours — stop and get help into the room, and secure the airway. Intubation, tracheostomy without prior intubation, a laryngeal mask, compression at the perforation, a monitored bed, transfer with theatre and ITU standing by, incision and drainage, extraoral exploration and vessel ligation, external carotid ligation, angiographic embolisation, and the implant's fate all belong to somebody else, later, in a building with an anaesthetist in it. Knowing which is which is the skill; doing the unlit ones yourself is the mistake. Reversing the anticoagulant is the dashed road — this corpus does not source it, and she is not on one. BARRIENTOS-LEZCANO, VERBATIM, if you want an authority: “Airway preservation is mandatory. The surgical approach for the bleeding source will depend on hematoma progression.” AND THE IMPLANT: where Law's cases recorded its fate, 9 of 17 stayed in, 5 were removed, and in 3 the placement was abandoned — the implant is almost never the emergency. Then say the last thing before you advance: nothing on this slide tells you WHY it happened, and the vote is worthless if anyone thinks they have already been told the cause.
how far back
today · nothing revealed yet
Walk-back 2 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe drill went through the lingual plate—
BShe was anticoagulated and nobody checked—
CThe implant was angled lingually at the last moment—
DThe mandible was never looked at in cross-section—
Hold on to what you picked. Two of those four are the same answer one
rung apart. One is not in this history at all — and the fourth is
where this walk-back ends.
+2:15. Chase abstainers — commitment is the whole mechanism. The bars are hidden until you advance one step, so nobody anchors on the early votes. Read the distribution out and say nothing about it. Poll id w2. If the phones cannot reach the laptop (NHS guest wifi very often isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. SAY THE RULE FOR THIS VOTE, because it is no longer on the slide: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — those are different questions and only one of them is answerable. An M&M answers the first, arrives at 'bad surgery', and stops. Three of these four are things that happen in mandibles; only one is a decision somebody made. THE EVIDENCE FOR VOTING FIRST, if anyone asks why we do this: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm. Expect a heavy vote for A. A is correct and it is useless: it names the hole, not the reason the hole was where nobody expected it. Do not say so yet.
how far back
before the flap · the shape she arrived with
Step back two · before the flap ·
walking past a rung — the perforation
The undercut is not a variant. At the second molar it is
the normal shape.
A · Levingston 2025Type Iunder 2 mmA · Levingston 2025Type II2 to 3 mmA · Levingston 2025Type IIIover 3 mm
Panels cropped from figure 2 and classified by fossa depth.
Levingston et al. 2025. Reproduced under CC BY 4.0. A morphometric CBCT series of 160 patients — not this
patient.
Shallow, moderate, deep. You cannot drill down the long axis
of the third.
200 of 200
A lingual concavity between the
first and second molar roots — mean depth 3.72 mm
Sen 2025, 200 CBCTs
90%
Lingual undercut at the second
molar — 56% at the first
Nickenig 2015
A lingual concavity is not a finding. It is the anatomy.
Observed a lingual concavity in 200 of 200 scans. Inferred how deep hers is — and the two CBCT series do not measure the same line.
+4:00. THE PIVOT. Put the three panels up and let the room read left to right before you speak — 1.92, 2.79, 3.22 mm on the figure's own annotations, and the third one is a shape you cannot drill down the long axis of. The line that matters is 200 of 200: a lingual concavity is not a finding, it is the anatomy. THE REST OF NICKENIG, which used to be a chart here: 56% at the first molar, 90% at the second, 68% across the posterior mandible overall. Do NOT use '66% of posterior mandibles' — that was Chan 2011 and it was edentulous first-molar sites only, and somebody in the room will quote it at you. LEVINGSTON'S OTHER NUMBER, if you want it: 6.2% of male right sides had the deepest point of the fossa BELOW the canal at the first molar — which is to say it is at or above it almost every time, at the depth you are drilling to. THE AMBER LINE IS WORTH A SENTENCE OF ITS OWN. Sen measured the concavity horizontally, from a line dropped off the mylohyoid ridge, and got a mean of 3.72 mm, which on Parnia's thresholds — the ones Sen cites — is the HIGHEST risk class. Levingston measured it perpendicular to a chord across the lingual cortex, applied the same thresholds, and found the SHALLOWEST class commonest. They are not measuring the same line, and Sen says so about the older studies in as many words. Take the classification for what it is: a way of describing one patient's scan, not a population risk you can quote at a coroner. A registrar who notices that has learned more than one who memorises either number. Do NOT quote this paper's concavity angles — 41.2 degrees is an annotation on one illustrative figure and two of its SDs are internally impossible; the project has ruled them out and they are on no slide.
how far back
the plan · before anything was cut
Step back three · the plan ·
the picture that was in your head
One of these is the view you had. The other is the same mandible.
B · Sen 2025Panel A · the sagittal planeheight, and nothing elseB · Sen 2025Panel B · the same mandibleridge, concavity, canal
Panels A and B of figure 1, cropped apart. Sen et al. 2025. Reproduced under CC BY. A
morphometric CBCT series of 200 patients — not this patient. Panel A is a
reconstruction rather than a panoramic film: the point is the plane, not
the machine.
The two hazards in this site are millimetres apart and they lie
in different planes. You measured one.
Panel A is every image most posterior mandibular implants are planned from.
It carries nothing about the wall 84% of these bleeds
come through.
Observed a plan with a height, a width, and a length. Inferred that the ridge was a box. It is why 13 mm felt like enough.
+5:45. Put both panels up and say nothing for five seconds. Then ask the room which one they could draw from memory for their last posterior mandibular implant — the honest answer is A, and the silence is the teaching. Be precise about the caveat in small print: panel A is a CBCT reconstruction, not a panoramic film. The argument is about the plane you reason in, not about which machine took the picture. SEN'S TWO DISTANCES, which used to be a table here: the mean distance from the deepest point of the concavity to the mandibular canal is 3.72 mm (range 2.03 to 6.56), and the mean depth of the mylohyoid ridge below the crest is 6.87 mm — the point below which the lingual plate starts falling away from you. The two hazards are millimetres apart and they lie in different planes, and you measured one of them. A ridge that is 7 mm wide at the crest can be 3 mm wide at the apex of your implant; height is the one dimension a panoramic gives you honestly, and it is the dimension that was never going to hurt her. THIS RUNG HAS SWALLOWED A WHOLE RUNG — the perforation — and here is what it said, in the order to say it. 21 of Law's 25 cases were caused by perforation of the lingual cortex; that is the 84% on the slide. The sublingual artery is the main supply to the floor of the mouth, and the submental artery frequently sends a branch there and is sometimes the ONLY supply — which is why Barrientos-Lezcano notes that a blind ligation of the lingual artery can fail outright, and that some of these arteries run parallel to the occlusal plane and lateral to the sublingual gland, where a drill is far likelier to divide them. Observed is the perforation, on the post-event scan, in the cases that got one; inferred is WHICH vessel, because in most published cases nobody ever sees it. The cases that get written up cluster in the interforaminal mandible while the concavity is deepest posteriorly — we do not have a site breakdown of Law's 25 and we are not going to guess one. And the flap: of the 13 roads on the soft-tissue fan, the 5 that were taken all keep the lingual plate under intact periosteum. The one road that would have put your eye on it — lingual reflection with a retractor on the plate — is also the one that brings the lingual nerve into the field. 'Always raise a lingual flap' is not the lesson and nobody should leave thinking it is. Concede the point to whoever voted for the drill going through the plate, in one sentence, and then change the question underneath them: not whether the plate was perforated, but why the plate was where nobody expected it.
how far back
weeks earlier · the plan, on a panoramic
Step back four · the plan, weeks earlier ·
the last rung, and the shortest
One in 25 had a scan.
The margin that was checked was vertical. Every margin that
mattered here was horizontal.
1 in 25
“Only 1 case involved preoperative 3-dimensional
imaging before implant insertion”
Law 2017
verbatim
“Preoperative 3D imaging helps to
visualize the individual mandibular shape, which could decrease the
incidence of serious complications”
Law 2017, conclusion
Observed 24 of 25 published cases had no cross-sectional imaging. Inferred that a scan would have prevented them — no comparison group, no denominator, no trial.
+7:15. THIS IS THE ARRIVAL. Say the number, wait, then read Law's conclusion verbatim off the slide — it is one of the very few sentences in the whole hour where an author states the fix, and the room should hear it in the author's words rather than yours. One scan, in 25 cases, across twenty-six years of English-language literature: 4%, and that 4% is the whole walk-back. It is also the only rung where the answer costs a scan rather than a skill. THEN SPEND AS LONG ON THE AMBER LINE AS ON THE NUMBER: 24 of 25 is not an effect size. There is no comparison group, no denominator and no trial, most posterior mandibular implants are still placed on a panoramic, and almost none of them do this — a registrar who leaves quoting 24 of 25 as an effect size has learned the wrong thing from the right slide.
how far back
weeks earlier · the plan, on a panoramic
12 roads out of this junction. 4 taken.
The imaging that was ordered
Taken — 4 of 12
Clinical examination and study castsPeriapical imagingPanoramic imagingSafety margin — 2 mm to the IAN
Available, not taken
Diagnostic wax-upCBCT — field of view and doseRadiographic stent and dual scanDICOM and STL alignmentSafety margin — mental foramen and anterior loopSafety margin — lingual concavitySafety margin — maxillary sinusSafety margin — nasopalatine canal
Thirty seconds. 12 roads, 4 taken — clinical examination and casts, a periapical, a panoramic, and one safety margin. Point at the lit chips first, then at the count. Two of the five safety-margin stations are maxillary and were never in play here; of the three that were, one was checked, and it was the vertical one. The lingual concavity is a road on this junction and nobody stopped at it — that is the sentence to land. AND THE 2 mm MARGIN ITSELF: this project's own foundations tree marks it unsourced, 'convention as much as evidence'. It is on the map because everyone uses it, not because anyone has shown it is the right number, and several people in the room will have been taught it as a rule. Say that out loud.
how far back
the plan · and the vote you cast before you saw it
The ballot, answered · walk-back 2 of 9
You voted before you were told anything. The walk ended at D.
AThe drill went through the lingual plate
BShe was anticoagulated and nobody checked
CThe implant was angled lingually at the last moment
DThe mandible was never looked at in cross-sectionwhere the walk ended
The plate was perforated — true, and useless: it names the hole, not the reason the hole was where nobody expected it. The concavity was in 200 of 200 scans, and the plan was made in the one plane that cannot show it.
Observed a perforated lingual plate, and a lingual concavity in 200 of 200 scans. Inferred that the drill was the error, rather than the plan.
Twenty seconds. If the poll server has been running, read this room's split off the bars on the ballot slide before you name anything — say the distribution out loud. Then name the option: the mandible was never looked at in cross-section. Most rooms put their weight on the drill going through the lingual plate, which is TRUE and is why it is worth ten seconds: concede it, then say that it names the hole and not the reason the hole was where nobody expected it. Do not reopen the anatomy here — the principle is the next slide and it is seven words. Move.
how far back
weeks earlier · before the plan was made
The floor · foundation 8 of 8
· anatomy, and its limits
The measurement was vertical. The
danger was horizontal.
Below the mylohyoid ridge the lingual wall turns away from you,
towards a gland, a fossa and two arteries.
Observed the danger in this jaw is lingual, horizontal, and below the ridge. Inferred that anyone failed — they almost all survive, and the implant usually stays.
Two of eight lit, and
one foundation from ten minutes of walking — the
shortest walk so far ends furthest from the drill.
+8:45. Let the headline sit before you speak — it is seven words and it generalises to the whole of foundation 8, which is the ONLY foundation this walk-back lights. It is the only one of the nine that lights a single card, and that is worth naming: this complication has one cause and it is anatomical. Then advance once and let the board fill under it. The posterior mandible is not a ridge with a height, it is a shape, and every routine image you take of it — periapical, panoramic, the mental picture you carry from the last one — measures the one axis that shape does not vary in. The danger is lingual, horizontal and below the ridge; everything you routinely measure is vertical, or buccal-to-lingual at the crest. That is the gap this walk-back walked back through. READ THE AMBER LINE PROPERLY, because the room has just been frightened for eight minutes and needs to leave with a proportionate picture. Barrientos-Lezcano's patient had a tracheostomy on the day, the tube out on day three, went home on day four, and the haematoma resolved in the second week; at one year the only sequel was a scar on the neck, no implant was removed, and he completed his rehabilitation. In Law's series 9 of 17 implants stayed where they were. So this is not an argument that the surgeon failed. It is an argument that the outcome stopped being avoidable WEEKS EARLIER, on the day somebody decided which picture of this mandible was enough — and the reason to take the scan is not fear of a death, it is that you cannot plan in a plane the danger does not lie in. WHAT CHANGES ON FRIDAY. Three things, none of them 'be careful', and do not cut these if you are running late. ONE: before a posterior mandibular implant, say out loud what the lingual plate does below the mylohyoid ridge at that site — if you cannot, you do not have the image you need. TWO: write the escalation on the plan, not in your head — the number you ring, where the airway kit is, and which hospital takes it. THREE: if the floor of the mouth is rising, the airway comes before the diagnosis, and you do not incise it to let it out. ASK THE ROOM to predict which card lights next; that prediction is the learning. Then hand over: six weeks, and it came out in her hand.
how far back
today · six weeks after placement
Walk-back 3 of 9 · UL1 ·
a composite patient — two published cases
She is 55. You placed this implant six weeks ago. Today it lifted out.
A · McCrea 2017“The implant has been simply ‘unscrewed’ with
minimal removal torque applied via Adams Orthodontic Pliers.” —
figure 8, the paper’s own caption. A different published case
from the one narrated: there the fixture had been in place 22 months and was
carrying a bridge. McCrea et al. 2017. Reproduced under CC BY.
One implant at UL1. Single unit, no graft. Stable when you seated
it — you remember the torque. She came in because the temporary felt loose.
Observed a fixture out under hand torque, and thread marks in the socket wall. Inferred nothing yet — not the bone, not the plan, not how long.
Single unit, no graft, good torque at seating. Every recorded step of
this was normal.
+0:00. Put the photograph up in silence and READ THE FIGURE CAPTION OUT LOUD, WORD FOR WORD — an implant removed with Adams orthodontic pliers. That sentence IS the walk-back, and it was written by somebody else, which is why it carries. Then give the history in the order written and STOP. Resist 'and of course you can see that...' — you cannot see anything, and that is the whole point. THE SEAM, said in thirty seconds before the photograph goes up (it is no longer a slide, because walk-back 2's scoreboard is the seam): two walk-backs so far, and both ended in an anatomy you could photograph — thirty seconds ended in a sinus, ten minutes ended in a lingual plate, and both times the thing that went wrong had a shape and a picture of it exists. This one is different and it is the reason the format has to work. Osseointegration is invisible when it holds and invisible when it fails: no radiograph of it, no probing depth for it, no number you can take in clinic. It has exactly one photograph and you only get to take it after the implant is already out. Two of the eight foundations are lit so far; tell them to decide privately which one this is going to be, and do NOT name it yourself. If the room hears 'osseointegration' now, the vote is worth less. THE REST OF THE HISTORY, if they ask: no membrane; no pain, no swelling and no discharge; she came back at two weeks and the tissue looked normal.
how far back
today · nothing revealed yet
Walk-back 3 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe plan never showed what was underneath the ridge—
BThe bone was overheated when the osteotomy was cut—
CIt was loaded before secondary stability had arrived—
DPart of the implant was never in bone at all—
Two of those four are the same answer one step apart. One of them is
where this walk-back ends — and it is not the one most rooms pick.
+1:30. Chase abstainers — commitment is the whole mechanism. The bars stay hidden until you advance one step, so nobody anchors. Read the distribution out and say nothing about it. Poll id w3. If the phones cannot reach the laptop (NHS guest wifi isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect the OVERHEATING and EARLY-LOADING options to take most of the room — name them, not their letters, because the spine permutes the ballot and the letters move between walk-backs. They are the two answers a surgical training teaches you to give, and neither of them is what happened here. THE RULE FOR THIS VOTE, said aloud because it is no longer a box on the slide: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — those are different questions and only one of them is answerable. An M&M answers the first, arrives at 'bad surgery', and stops; we keep walking past that. THE EVIDENCE FOR VOTING FIRST, in one sentence if you want it: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm.
how far back
six weeks · the reviews that showed nothing
Step back one · six weeks · every review appointment
in between
There was never a moment when this looked like it was failing.
A · McCrea 2017The film“no obvious signs”A · McCrea 2017The flapno bone on the surface
Figures 2 (left film only) and 7, cropped. A different
published case from the one narrated — same site, UL1. McCrea et al. 2017. Reproduced under CC BY.
Removal torque is the only measurement anyone ever makes of this
interface — and it is made once, at the end.
Six weeks of appointments, and not one of them could have caught
this. There is no chairside test for what had gone wrong.
Observed a film read as normal, and no bone on the surface at the flap. Inferred that it never integrated, rather than integrated and then failed.
+3:30. The pivot of the first half. Hold on the left-hand film for longer than feels comfortable and ask the room what is wrong with it. Nothing is. That is the answer. THE TWO EVIDENCE ROWS, now off the slide: that radiograph was reported as 'no obvious signs of peri-implantitis' at 22 months, on a fixture whose apical portion had never been in bone and which then unscrewed by hand (McCrea 2017, fig 2); and how often does this happen — this project has NO verified denominator for early failure to integrate, so there is no number here rather than an invented one. Say that honestly; a made-up rate would be worse than none. The sentence to land: a fixture that never integrated can look, on the only image you routinely take, exactly like one that did — foundation 1 is the only foundation in this hour with no sign of its own. THE WHOLE OF THE OLD 'OTHER WAYS TO LOSE IT' RUNG LIVES HERE NOW, because it covered these same six weeks, and anyone who voted A or B is owed it. There are six ways to lose integration and none of them is what happened here: overheating the osteotomy, micromotion beyond the threshold, early infection, loading before secondary stability, smoking, irradiated bone (11-foundations-tree.md §1.5). On overheating — VERIFIED, and it is the uncomfortable one: guide sleeves impair irrigation and raise bone temperature (Ashry 2021; Sannino & Gherlone 2018; Orgev 2020; nuance in Tur 2025). The thing that makes you more accurate is also the thing standing between your irrigant and the bur. Overheating is not a beginner's error you grow out of — it is a cost you can buy accidentally by choosing the more careful technique: you buy position with heat, and you pay for it at an interface nobody can inspect, weeks after everyone has gone home. On loading — immediate versus delayed loading of single crowns shows no significant difference in satisfaction, quality of life, biological complications, mechanical complications or survival (Ribeiro 2024, 20 studies); the one subgroup that moved was immediate implants in the posterior mandible, with significantly higher marginal bone loss (p = 0.001, Ribeiro 2024); and immediate placement with provisionalisation in the aesthetic zone showed no crestal bone or papillary difference but implant-related complications more often (Gaddale 2024, 16 studies). So loading protocol is not where the risk is: survival is equivalent, what changes is the complication rate and, in one site, the bone level — which makes immediate loading a STABILITY decision, foundation 2, not this one. AND THE PICTURE WE CANNOT SHOW YOU: there is no obtainable CC BY photograph of thermal necrosis in this project's licence-checked library. Say so; do not apologise for it and do not substitute something that is not it. Naming the gap is the teaching.
how far back
six weeks earlier · during the osteotomy
Step back three · six weeks earlier ·
the second the drill went past the bone
Part of it was never in bone. It was in a hole already there.
A · McCrea 2017The probeinto a canal, from inside the osteotomyA · McCrea 2017Debridedthe conduit, patent
Figures 9 and 10. The paper’s captions: probing “to
demonstrate that the apical portion was in the nonresistant nasopalatine
canal”, and the graft “easily debrided… to display the very
patent aberrant canal”. McCrea et al. 2017. Reproduced under CC BY.
No surface and no alloy would have changed this. The apex was in a canal,
and no surface integrates with air.
Bone integrates with what it touches, and with nothing at all
that it does not.
Observed a probe passing without resistance into a canal from inside the implant bed. Inferred that this is why it never integrated — the author infers it too.
+7:00. The answer arrives. Whoever voted that part of it was never in bone was right one step early — say so, and then keep walking, because that is a finding and not yet a cause. THE CHAIN, said rather than drawn: the osteotomy was taken to depth in an anterior maxilla treated as a solid block of bone; the apex entered the nasopalatine canal, which is soft tissue and vessel, not bone, so there was nothing to integrate against; no interface ever formed at the apical third, and what interface there was could not hold the rest. There is no partial credit at this interface — an implant two thirds in bone is not two thirds integrated, it is an implant with a lever buried inside the jaw. The amber line matters here: this is the ONE rung in the whole walk-back where the inference is close to demonstration, because a probe went somewhere a probe cannot go through bone, and the room should feel the difference between this and the softer inferences on either side of it. The roads out of 'which implant' are the slide after this one — do not start on surfaces here, the sentence about air is the end of this rung.
how far back
six weeks earlier · the implant on the tray
13 roads out of this junction. 2 taken.
The implant that was chosen
Taken — 2 of 13
Commercially pure titanium, grades 1–4Surface — sandblasted, large-grit, acid-etched
Thirty seconds, then the two sentences that matter. Point at the two lit chips first — commercially pure titanium and a sandblasted, acid-etched surface. NEITHER PUBLISHED CASE RECORDS ITS SURFACE: this is the composite, and it is the commonest answer in the UK. Say that before anyone asks. One: Three of the dashed chips are dashed because the field ABANDONED them — machined, titanium plasma-sprayed, hydroxyapatite plasma-sprayed — and two more because they are not evidenced; every one of the abandoned ones was, in its decade, the surface that changed everything, and you will be shown the next one before Christmas. Two: the strongest consensus review in this day's evidence base can find the POSITION in the outcome data and cannot find the surface — of every surgical and implant-related factor examined, inadequate implant position is the one clinical evidence proves predisposes to peri-implantitis, while implant-related factors had insufficient evidence to draw associations (Monje 2025, AO/AAP consensus SR, 34 articles). The foundations tree carries exactly one warning on this concept and it is not a biological one: surface generation claims are heavily industry-funded — check who paid. If a registrar names a specific product here, ask them who funded the trial they are quoting. That is the exercise.
how far back
before the drill was ordered
Step back five · before anything was ordered ·
the last rung where the answer could still be no
The canal was not on the plan. In both published cases it was on the scan.
A · McCrea 2017On a plain filman orifice at the apex of the implantB · Volberg 2019A second casepalatal necrosis after removal
Left: figure 3 — the canal orifices “seen
clearly… one at the apex of the implant”. McCrea et al. 2017. Reproduced under CC BY. Right: a second
published case — canalis sinuosus damage at UL2. Volberg et al. 2019. Reproduced under CC BY.
A scan you do not read is a panoramic. The rung is not “take a
CBCT” — it is: name the structures you intend to avoid, in writing,
before anyone orders a drill.
“During the preoperative CBCT evaluation and dental implant
planning, none of the nerve structures around left lateral incisor were
noticed” — Volberg 2019.
Observed two anterior maxillae, two preoperative CBCTs, the structure visible on both in retrospect. Inferred that reading the scan would have prevented it. Neither author shows that.
+11:00. Do not let this become an advert for CBCT. The WHOLE FORCE of the rung is that both published cases HAD one. Ask the room, by hands, who writes the nasopalatine canal into their operation note before the day of surgery. Then read Volberg's recommendation out and let it sit: an author who has just published a complication recommends not placing the implant — defects at risk here should be restored WITHOUT an implant, or FULLY GUIDED. Both ends of the planning line, and nothing in the middle. THE NUMBER, if challenged: freehand placement deviated up to 6.3°, and the worst apex-to-nerve error was 2.55 mm freehand versus 0.63 mm fully guided — in vitro, on printed mandibles, so read it as a statement about freehand and not about this site (Mistry 2021). THE PLANNING FAN is off this slide (the grammar allows one fan on screen and the coverage audit assigns it to the implant row) but it is in the print appendix and it is worth naming out loud, west terminus to east: look in the mouth — clinical examination and study casts, diagnostic wax-up, periapical and panoramic imaging, CBCT with its field of view and dose, DICOM and STL alignment, prosthetically-driven planning, a safety margin to the nasopalatine canal — then constrain the drill: freehand, pilot-drill guide, partially guided, fully guided static, the guide sleeve with its irrigation and thermal cost, dynamic navigation, and two dashed chips (robotic, AI-assisted) that are there because they exist and not because the evidence puts them here. This case took the third road and the eighth: plain films, then freehand.
how far back
before the drill was ordered · and the vote you cast first
The ballot, answered · walk-back 3 of 9
You voted before you were told anything. The walk ended at A.
AThe plan never showed what was underneath the ridgewhere the walk ended
BThe bone was overheated when the osteotomy was cut
CIt was loaded before secondary stability had arrived
DPart of the implant was never in bone at all
Overheating and early loading are the two answers a surgical training teaches you to give, and this case was neither. The apex sat in the nasopalatine canal from the first turn — no surface integrates with air — and that canal was never on the plan.
Observed an apex in the nasopalatine canal, and a fixture out under hand torque. Inferred that the bone was overheated, or the implant loaded too early.
Twenty seconds. If the poll server has been running, read this room's split off the bars on the ballot slide before you name anything — say the distribution out loud. Then name the option: the plan never showed what was underneath the ridge. Most rooms put their weight on overheating and early loading, and this is the moment to say why neither survived the walk — there are six ways to lose integration and this was none of them. Give the credit that is owed: whoever picked 'part of it was never in bone' was right one rung early. Then move — the principle is the next slide.
how far back
before the plan was made
The floor · foundation 1 of 8 ·
osseointegration
Bone tolerates it. It never bonds to it.
Not adhesion, and not a weld. An equilibrium — living
bone holding still against a foreign body it has agreed to wall off instead of
reject. You can only avoid preventing it, and every way of preventing it is
a decision taken before the fixture is opened.
Observed a fixture removed under hand torque, and the symptoms resolving after removal. Inferred that it never integrated — nobody took a biopsy.
Three of eight lit, and the walk was six weeks long. Six walk-backs to go.
+15:00. Let the headline sit before you speak. Everything else in this hour is protecting this one line, so say it slowly and do not soften it: bone does not bond to titanium, it tolerates it. Equilibria have to be allowed to form before they can be maintained, and they form ONLY where bone touches metal — where the osteotomy left the bone, there was never anything to maintain. If you add one sentence, add that osseointegration is the only foundation you cannot inspect, cannot measure in clinic and cannot repair; you can only decide, in advance, not to prevent it. AND THE HALF THAT KEEPS IT HONEST, which is not on the slide and should be said: in the published case the implant was removed, the perforated canal was covered with a membrane, the defect was grafted, and the nasal bleeding and sinusitis were arrested, the sense of blockage disappeared 'immediately', and the site remained symptomless. Of everything in this hour a failure to integrate is among the most RECOVERABLE — nothing was lost except the implant and the time. Say that out loud, because the walk-back is not an argument that this surgeon was careless; it is an argument that the outcome stopped being avoidable at the plan, weeks before anyone was in theatre. (The immediacy of that resolution is the strongest thing the author has, and he says so.) Then advance and let the board fill: ask the room to predict which cards light next — if they can, the format is working. WHAT CHANGES ON FRIDAY. Three things, none of them 'be careful'. One: in the anterior maxilla, write the nasopalatine canal and the canalis sinuosus into the plan by name, with a distance to your intended apex — a structure you have not named is a structure you have not avoided. Two: record insertion torque and the date, every time; it is the only measurement anybody ever takes of this interface and the one thing you will wish you had written down. Three: if you drill through a sleeve, treat cooling as a separate problem you have just acquired. CLOSE ON THE FORMAT CLAIM: foundation 1 has no photograph of its own, no radiographic sign, no probing measure and no number in this project's evidence base for how often it fails early — and it was reached in six slides, through the one photograph that exists of it, taken thirty seconds after it stopped mattering. Then hand over: seven months, and the membrane is showing. The walk gets longer from here and the answers stop being anatomical.
how far back
Definitions · foundation 1 of eight ·
it has been earned; now name it
Osseointegration
OsseointegrationA stable, direct structural and functional connection between living bone and the surface of a load-bearing implant.
Bone-to-implant contactThe proportion of an implant surface touching bone in a histological section. It is not a clinical test.
Foreign-body equilibriumA stable state in which bone contains and tolerates the implant rather than eliminating it.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · seven months after the graft
Walk-back 4 of 9 · UR3 to UR5 ·
a composite patient — one published case
She is 52. Seven months ago you grafted this ridge for three implants.
+0:00. THE SEAM SLIDE IS GONE and it does not come back — walk-back 3's scoreboard was the seam. Say this in one breath while the four panels are going up, and put nothing between it and the ballot: three walk-backs, and not one of them ended at the drill; they ended in a plan, in a panoramic, and in a site that was never going to hold an implant. This one ends somewhere new — in the last five minutes of an operation that had otherwise gone correctly, and then one step further, in whether the operation needed to happen at all. Three of the eight foundations are lit so far; tell them to decide privately which one this is going to be, and do NOT name it. Then STOP TALKING and let the room read left to right. The time series does the work — nobody needs telling that it is getting worse. NOT ON THE SLIDE, give it in this order and then stop: you extracted, decorticated, tented the site with screws, packed it, covered it with a titanium-reinforced d-PTFE membrane and closed it; she has been seen every month since; she has had two courses of co-amoxiclav and she rinses with chlorhexidine. Every extra fact you add makes the vote on the next slide worth less. THE FULL AMBER LINE, if you want it: what is NOT yet known is what is under the membrane, how much bone is there, how the flap was closed, and whether she needed a graft at all.
how far back
today · nothing revealed yet
Walk-back 4 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe flap was closed under tension and it broke down—
BWrong membrane for a site with this little soft tissue—
CThe graft was never necessary — the whole operation was avoidable—
DContamination tracked in under the edge of the membrane—
Hold on to what you picked. Two of those four are the same answer one
step apart — and a third is where this walk-back actually ends.
Scan once — it follows the deck
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+1:30. Chase abstainers — commitment is the whole mechanism. The bars are hidden until you advance one step, so nobody anchors on the early votes. Read the distribution out and say nothing about it. Poll id w4. If the phones cannot reach the laptop (NHS guest wifi very often isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect a heavy vote for the tension answer: it is what registrars are trained to say, it is half right, and it is not specific enough to act on. NOT ON THE SLIDE, and only if someone asks why they vote before being taught: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT) — watching is the losing arm. SAY ALOUD, because it is the rule for this vote and it is no longer printed: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable. An M&M answers the first, arrives at 'bad surgery', and stops. We keep walking past that.
how far back
today · the membrane is off
Step back one · the membrane comes off, fourteen days after
the pus · one rung walked past — how common this is
The same patient exposed twice. Only one of them suppurated.
A · Heggendorn 2022The membrane, outtexture changed at the exposure siteA · Heggendorn 2022Under itgranulomatous tissue, curetted and sent
Panels cropped from figure 5. Heggendorn et al. 2022. Reproduced under CC BY.
The first exposure lifted at the lateral margin, and fluid
tracked underneath into the space the graft was meant to fill.
Observed two exposures in one mouth — one at the margin, one central and still covered. Inferred that the infection destroyed the graft; the authors write “possibly”.
The barrier is not the seal. The peripheral contact between barrier and
host bone is the seal — protect the margins, not the middle.
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+3:30. This is the mechanism slide and the one thing they should be able to repeat back on Friday: protect the margins, not the middle. THE RUNG IN FRONT OF THIS ONE IS CUT — say it in three sentences before you show the photographs, because the room has to stop blaming somebody before it can hear a mechanism. One: between one in 3 and one in 5 of these expose, and exposure of the barrier or the block graft is the most common complication of oral regenerative surgery (Sanz-Sánchez 2022; Urban 2023 — 'the most common postoperative complications are wound dehiscences'). Two: this membrane is d-PTFE, its pores are about 0.2 µm, smaller than the bacteria, so unlike e-PTFE it does not require primary closure and is DESIGNED to tolerate being exposed — Laurito left ten of them deliberately open for 28 days and got plaque and marginal inflammation, no infection and no pain in any of them. Three: therefore 'it exposed' is not the diagnosis. The question is not whether the membrane showed, it is WHICH PART of it showed — and that is this slide. THE NUMBERS, if challenged (5 independent series): Nan 2023 32.8% (prospective, 61 augmented sites); Ragucci 2025 30.9% (meta-analysis · conventional titanium mesh); Ng 2025 25.0% (umbrella review of 8 systematic reviews); Soares 2025 22.2% (9 cases · complete early exposure — both lost the graft); Ragucci 2025 20.3% (meta-analysis · CAD/CAM titanium mesh). Also Sabri 2024, 22 articles — pooled complication rate 10.8%, mesh exposure commonest and graft failure second; Sánchez-Sánchez 2021 — 3.2 to 13.6% for the Urban technique; Soares reports 7 of 9 exposed at some point. THE CHAIN, spell it out over the two photographs: lateral edge lifted at the mesial margin at three months, fluid and contamination tracked underneath, then infection and purulent exudate, and on the scan a void between membrane and crest at UR4 — the graft had gone. Nan 2023: exposure correlated with reduced bone gain in every dimension, p < 0.001, and severe vertical defects were significantly more likely to expose. THEN REFUSE TO MAKE IT TIDY, and this is why the second photograph is here: the histology showed chronic inflammation and neutrophils, yes, but also CALCIFICATIONS. Bone was still forming underneath an infected membrane that was about to be thrown away. Do not let the room settle on 'infection ruins grafts'.
how far back
seven months earlier · forty minutes before closing
Step back two · the same operation, forty minutes earlier
· one rung walked past — the closure
Two decisions were taken here. Neither of them is the one that failed.
A · Heggendorn 2022What went infig 3D — “insertion of the graft”A · Heggendorn 2022What covered itfig 3E — “coating with the L-PRF membrane”
Panel letters and quoted wording are figure 3’s own
caption. Heggendorn et al. 2022. Reproduced under CC BY.
Regeneration was predictable irrespective of the
biomaterial, whenever the biological principle was followed.
Observed a xenograft and an alloplast, under a titanium-reinforced membrane, with L-PRF over the top. Inferred that any one of those three choices is what failed.
Only the autografts are osteogenic; everything else here is a scaffold
— and where bone is harvested the donor pays: 64.5% back at six
months, 89.2% at a year.
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+6:30. Stay on the two photographs; the graft taxonomy is the NEXT slide, on its own, and it has thirty seconds of its own. THE BARRIER TAXONOMY IS NOT ON SCREEN AT ALL. Name it instead, in one sweep, and say the last four are dashed because they exist rather than because the evidence puts them here — 15 ways to cover a graft: Titanium-reinforced d-PTFE, Plain d-PTFE, e-PTFE, Native collagen, Cross-linked collagen, Titanium mesh, Customised CAD/CAM mesh, Pericardium / amnion-chorion, Tenting screws, No barrier at all, L-PRF over the top, PRP / PRGF, Enamel matrix derivative, rhBMP-2, rhPDGF. Biologics are the least sourced concept in this whole day. Calciolari 2023 (network MA) also found complications HIGHER with cross-linked collagen membranes, which is the one material choice on that list that does change the odds. THE RUNG BEFORE THIS ONE IS CUT — the closure — and it is the anti-blame rung, so do not skip it: the registrars will want to convict the operator, and the amber discipline is the brake. Nothing in the operation went wrong until the flap had to lie down. The paper's own two candidate mechanisms, verbatim: 'insufficient flap release leading to lack of passivity, with increased tension and damage to the suture', or soft-tissue necrosis caused by the periosteal releasing incision itself. OBSERVED: a closed flap, and a dehiscence three months later. INFERRED: that the closure was under tension — nobody measured it, nobody photographed it, and the authors of this case are inferring it exactly as we are. It is the most plausible explanation on the table and it is still an inference. Sanz-Sánchez 2022 associates exposure with flap elevation and a tensionless closure, and with patient selection (non-smokers lower). PASS — primary closure, angiogenesis, space maintenance, stability — and the first of the four is the one you do last, when you are tired. THE DECISION RULE THEY CAN USE: they re-sutured the dehiscence twice, both times it re-opened within seven days, and after the second operation the new membrane was exposed in ONE WEEK, because the tissue that failed you the first time is now scar. Re-suturing a dehiscence is not a rescue, it is a delay, and every operation you do here makes the next closure worse.
how far back
seven months earlier · forty minutes before closing
14 roads out of this junction. 2 taken.
What went in the defect — every graft material that was on the trolley that morning, and the two that were used
Thirty seconds, and it is a slide rather than a footnote. Point at the two lit chips first — bovine xenograft and synthetic HA — then at the count above them. The junction exists because a walk-back is a causal instrument and cannot arrive at a list, and this is the only place in the whole day the graft taxonomy can live without stopping the story. Say the mechanism the chips are hiding: only the autografts are osteogenic, everything else on that board is a scaffold — osteoconduction, in some cases osteoinduction, and nothing more. Do not read the column out; the room can read it faster than you can say it.
how far back
seven months earlier · before the flap was raised
Step back three · before the flap was raised ·
the only rung where the answer can still be no
Two operations later, the implant that went in at UR5 was
5 mm long.
Short implants are non-inferior to sinus lift on survival, and
better on marginal bone loss and biological complications.
Observed a 4.0 × 5 mm implant went in at UR5, and it integrated. Inferred that it would have worked on day one — nobody randomised this ridge.
The last step is not close it more carefully. It is that the graft
is not the safe default — a choice with a complication rate, made
before the blade.
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+9:30. Two sentences and the consequence — the sixteen roads are the next slide and they get their own forty seconds, so do not spend them here. THE AMBER LINE IS DOING REAL WORK: the buccal plate at UR5 had already gone, so the counterfactual is not available — the junction is the teaching point, not the counterfactual. What IS available is that nobody wrote down the alternatives before grafting. Ask the room, by show of hands, who documents the non-graft option in their consent, then move on quickly. THE NUMBERS: Mester 2023, five RCTs at five years or more — survival RR 0.97, not significant; Alenezi 2025, seven RCTs, 393 patients, 474 implants — marginal bone loss 0.26 mm favouring short, p < 0.001; set that against a membrane-exposure rate of one in three at the top of the range. SAY ALOUD: this is rung 2 of the anchorage ladder — avoiding the graft — and they have just arrived at it backwards, from a photograph of pus, instead of forwards from a list of techniques. That is the whole method in one slide.
how far back
seven months earlier · before the flap was raised
16 roads out of this junction. 1 taken.
Roads out of a ridge that will not take a standard implant
Forty seconds, and the room does the talking. Point at the single lit chip first — graft, then a standard implant — and then say the count out loud: that is one road of sixteen, chosen before the blade. NOT ALL OF THEM FIT THIS SITE, and naming the ones that do is the exercise: ask the room which of the other fifteen were genuinely available at UR5, and take three answers, no more. Do not let it become a short-implant advert — short implants are one chip on that board, not the moral of the slide.
how far back
seven months earlier · before the flap was raised
The ballot, answered · walk-back 4 of 9
You voted before you were told anything. The walk ended at C.
AThe flap was closed under tension and it broke down
BWrong membrane for a site with this little soft tissue
CThe graft was never necessary — the whole operation was avoidablewhere the walk ended
DContamination tracked in under the edge of the membrane
Tension is the trained answer and it is half right — but nobody measured that closure, and d‑PTFE is designed to tolerate being exposed. Closing it more carefully was never the last free decision. Grafting at all was.
Observed a flap closed, and a dehiscence three months later. Inferred that the closure was under tension — nobody measured it.
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+11:00. Thirty seconds, and it is the debt the hour owes them for voting. Read the lit option out, then say which one the room actually chose — you have the distribution from the ballot slide. MOST OF THEM WILL HAVE PICKED THE TENSION ANSWER and they should hear, in one sentence, that it is not a stupid answer: it is the paper's own leading hypothesis and it is still an inference nobody measured. What makes it the wrong STOPPING POINT is that it cannot be acted on — 'close it better' is not a decision, it is a wish. The graft decision is a decision, it has a complication rate, and it is taken before the blade. DO NOT LET THIS BECOME A SCOLDING: nobody in the room has been shown to be wrong, they have been shown where the walk goes one rung further than the vote did. Then straight into the principle — do not pause between them.
how far back
seven months earlier · before the incision
The floor · foundation 3 of 8 ·
bone and graft healing
Regeneration is a space, not a substance.
A graft is mostly a scaffold. What makes it bone is whether the
space stays closed, still, vascular and sealed at its edges.
Primary closure, angiogenesis, space, stability — and
all four are spent in the last five minutes, not in the pot.
Observed this graft was infected, drained, removed early, and lost material at UR4. Inferred that the surgeon failed — four months on, the implants looked integrated.
Five of the eight lit, and the walk was 7 months long. Five walk-backs to go.
1Osseointegration1 walk-back
2Primary stability1 walk-back
3Bone and graft healingprimary here
4The soft tissue sealalso lights
5Load transfer without a PDLnot yet
6Three-dimensional positionnot yet
7Biofilm and the peri-implant lesionnot yet
8Anatomy and its limits3 walk-backs
← backwardsThe principleThe decision to graftWhat went inThe closureThe exposureThe outcome
+12:00. Let the headline sit before you speak; the board is already under it, so advance once and put the count on it. This should feel like an arrival, not a summary. If you say one more sentence than is on the slide, say this: the argument about which graft material to buy is the least important argument in this room. THE TWO FINDINGS THAT CARRY IT, and they are not on the slide: regeneration was predictable irrespective of biomaterial whenever the biological principle was followed (Calciolari 2023, network MA), and exposure costs bone gain in every dimension, p < 0.001 (Nan 2023, 61 sites). The material is where the arguments are; the space is where the outcome is. WHAT CHANGES ON FRIDAY — three things, none of them 'be careful', and this is the part to protect if you are running late. One: decide whether you are grafting before you raise the flap, and write in the notes what you would have done instead. Two: budget the closure before you start — if the flap will not lie passive over the membrane, the operation is already over and the rest is theatre. Three: protect the margins, not the middle. THE HALF THAT MAKES IT HONEST, say it out loud: this graft was infected, drained pus, was removed early and lost material at UR4 — and four months after the implants went in the radiograph suggested osseointegration, there was no mobility, and she was temporised. A complication is not a catastrophe, and this is not an argument that the surgeon failed. It is an argument that the outcome stopped being fully controllable at the closure, and that the decision which could have avoided the whole sequence was taken before the flap was raised. AND ONE LAST TIME: observed is one published case, one ridge, one operator, and a literature that agrees exposure is common and costs bone; inferred is that those three actions would have prevented this. A case report cannot show that, and neither can nine of them — what they can show is WHERE THE DECISIONS WERE. ASK THE ROOM to predict which card lights next; that prediction is the learning. Then hand over: eighteen months, and you can see the metal through her gum.
how far back
Definitions · foundation 3 of eight ·
it has been earned; now name it
Bone and graft healing
OsteogenesisLiving cells in the graft form new bone.
OsteoinductionSignals recruit host cells and direct them towards a bone-forming lineage.
OsteoconductionA scaffold permits vessels and new bone to grow through it.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
Definitions · foundation 4 of eight ·
it has been earned; now name it
The soft tissue seal
Soft-tissue sealThe epithelial and connective-tissue cuff separating oral biofilm from the bone around an implant.
Keratinised mucosaOral mucosa with a keratinised surface surrounding the implant.
Emergence profileThe shape of the abutment and crown as they pass from the implant platform through the soft tissue.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · eighteen months after placement
Walk-back 5 of 9 · anterior maxilla ·
a composite patient — three published cases
She is 54. She says it looks grey.
A · Wen 2024Labiala fistula, and the papillae have goneB · Sangkhamanee 2024Above the margin3 × 3 mm of nothingB · Sangkhamanee 2024Two weeks later4 × 4 mm — the fixture
Panels cropped from Wen figure 1 and Sangkhamanee figure 1.
Wen et al. 2024. Reproduced under CC BY. Sangkhamanee et al. 2024. Reproduced under CC BY.
One upper anterior tooth, fractured.
Extracted and implanted in the same appointment eighteen months ago. It
integrated, it has never been mobile, and the radiograph is unremarkable.
Observed a fistula, papillae 1.5 to 2 mm short, a hole in the mucosa, metal. Inferred nothing about the bone, the position, or the socket.
It integrated, it has never been mobile, the radiograph is unremarkable.
Everything usually measured here is normal.
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+0:00. No preamble — the hour is already running and this walk-back starts on the photograph. Put the three panels up in silence and count to ten before you read the history. Read it in the order written and then STOP. Do not say the word 'position' and do not say the word 'buccal'; both are answers. NOT ON THE SLIDE, and worth saying once she has been introduced: she has been brushing it exactly as she was told. That matters, because the first thing a room does with recession is blame the patient. THE AMBER LINE IN FULL, if you want it aloud: observed is a fistula, papillae 1.5 to 2 mm short of where they should be, a hole in the mucosa and metal in the mouth; NOT yet known is what the bone is doing, where the implant actually sits, how thick the tissue was to begin with, and what was done with the socket on the day.
how far back
today · nothing revealed yet
Walk-back 5 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe tissue was always too thin here — she needed a graft at the start—
BIt should never have gone in the day the tooth came out—
CThis is peri-implantitis — biofilm has got under the seal—
DThe implant is in the wrong place—
Hold on to what you picked. The first answer is the one this room is
trained to give — it is not wrong, and it is not where this
walk-back stops.
Scan once — it follows the deck
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+1:30. Chase abstainers — commitment is the whole mechanism. The bars stay hidden until you advance one step, so nobody anchors. Read the distribution out and say nothing about it. Poll id w5. If the phones cannot reach the laptop (NHS guest wifi isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect A to win comfortably. A is the soft tissue answer and this whole walk-back is an argument about what sits underneath it. Two of the other three are the same answer one step apart. NOT ON THE SLIDE ANY MORE, and both worth one sentence each: the rule for this vote is that you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — only one of those is answerable; and note what is not on the ballot, because nobody here was careless and every option in front of them is a decision somebody made for a reason. THE EVIDENCE FOR VOTING FIRST, if challenged: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm.
how far back
about a year back · before anyone looked underneath
Step back one · what the room just voted for
≥2 mm buys tissue health. It has never been shown to buy
an implant.
Plaque index5.34
Mucosal recession4.05
Gingival index3.13
Peri-implantitisAfrashtehfar 2023 · no keratinised mucosa vs some, GRADE moderate2.78
Marginal bone loss1.85
Equivalent odds ratios, <2 mm against ≥2 mm
— Sabri 2025, except the peri-implantitis row. Ravidà
2022 puts the risk contribution low; Zhang 2025 agrees on
direction. Both no-derivatives: numbers quoted, no figure.
Same review, no significant association with bleeding on probing, probing depth, implant survival, disease prevalence.
“≥2 mm buys you tissue health, not implant survival.”
Observed the association, now at moderate certainty. Inferred that grafting her tissue would tell you why the margin went.
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
MDT1.3AK1.3
Sabri 2025, umbrella review of ten systematic reviews · Afrashtehfar 2023, appraisal of a 22-study review doi:10.1038/s41432-023-00913-4
+3:30. This is the slide the room came for and the one they will misquote. Say the honest line twice, and say it in those words. THE RUNG WE WALKED PAST — 'you are not unlucky', and it is worth ninety seconds because it takes blame off the table before the walk starts: reported prevalence of peri-implant soft tissue dehiscence or defect, 4 figures — 16.9% Any implant, any site (Sangkhamanee 2024); 54.2% Aesthetic zone · patient level (Sangkhamanee 2024); 56.8% Aesthetic zone · implant level (Sangkhamanee 2024); 64.0% After an immediate, aesthetic zone (Wen 2024). The rate roughly triples between 'an implant' and 'an immediate implant in the front of the mouth', and Wen also reports that over a third of patients having an immediate placement in the aesthetic zone need soft tissue augmentation to get an acceptable result. DO NOT SOFTEN THE PROVENANCE CAVEAT: all four of those figures are quoted inside the two case reports this walk-back is built from, each citing an earlier series — we have read the case reports, we have not read the primary series, and that is weaker provenance than the rest of this hour. If someone challenges the 64%, agree with them: it is a CEILING figure from a case report's introduction. The teaching point is the gradient, not the decimal. At that rate this is not a complication of the operation, it is a property of the operation. WEN'S SIX CONTRIBUTING FACTORS, read the list and then count: implant malposition, labial bone fracture or fenestration, thin gingival biotype, absent or insufficient keratinised mucosa, local inflammation, an improper restoration profile. Two of the six are the soft tissue. The rest of this walk-back is the other four. THE NUMBERS BEHIND THE FOOT, if challenged: Ravidà 2022 ran a trial sequential analysis and found the impact of keratinised mucosa width as a risk factor for peri-implant disease 'remains low', with only plaque index significant; Zhang 2025 pooled 30 articles and found ≥2 mm associated with reduced plaque, gingival and bleeding indices and less bone loss. Neither may be reproduced — no-derivatives — and no figure of theirs is anywhere in this deck. SAY ALOUD, because it is off the slide: the foundations tree marks the ≥2 mm threshold 'widely quoted and weakly evidenced — check before teaching it', four reviews later it is better evidenced than that, and it is still not a survival threshold. A number repeated in every viva is not the same as a number that has been tested. THE SOFT TISSUE FAN IS NOW IN THE APPENDIX, not on screen — 14 roads at this junction, and the ones taken were: Nothing - at the time of placement, Connective tissue graft - envelope, Roll flap and pedicle techniques. Two of those three were done eighteen months late. Grafting her tissue would have made this cleaner, pinker and less recessed; it would not have told you why the margin went in the first place. Keep walking — the next slide is where the walk-back actually turns.
how far back
eighteen months back · and every month since
Step back two · the CBCT nobody had taken
The tissue is thin because there is nothing behind it.
B · Sangkhamanee 2024The scanlabial dehiscence to the apexC · Monje 2025In sectiona different case, same finding
Left: Sangkhamanee figure 1b — apical labial bony
dehiscence and 5.56 mm of vertical bone loss mesially and distally. Sangkhamanee et al. 2024. Reproduced under CC BY.
Right: Monje figure 2E(d), a different published case — the
deficient buccal bone under an implant placed in an ungrafted narrow ridge. Monje et al. 2025. Reproduced under CC BY.
Of every surgical factor examined,
implant position is the one the evidence proves predisposes to
peri-implantitis.
Observed a labial plate missing to the apex, and the hole sitting over it. Inferred that the hole is there because the bone is not.
In the Wen case the CBCT was normal and the margin still failed.
The authors put that one down to position too — too close to the
next tooth, and too shallow.
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+7:00. The turn. Everyone who blamed the thin tissue finds out here that they were describing the outcome rather than the cause. Give them the Wen counter-case in the crimson box before they can feel got at — a normal CBCT with a failed margin is real, and it stops this becoming a slogan. If you say one extra sentence, say: you cannot graft your way out of a position. THE NUMBERS, if challenged, because they are off the slide now: across 34 studies, of every surgical and implant-related factor examined, inadequate implant position is the one clinical evidence proves predisposes to peri-implantitis — and the review calls that evidence MODEST and SPARSELY REPORTED, so do not oversell it (Monje 2025, AO/AAP consensus SR). One included study puts malposition against peri-implantitis at OR 48.2, 95% CI 11.4 to 204.1 (Canullo, in Monje 2025) — a confidence interval that wide is a signal, not a measurement, and saying so is the evidence-literacy point. No implant surface, no connection and no thread form reached that bar; where you put it did. THE FULL OBSERVED LINE: in the Sangkhamanee case, a labial plate missing to the apex, a 6 mm pocket bleeding profusely, and a soft tissue hole sitting directly over it — and on the left-hand scan the reconstruction shows the fixture OUTSIDE the labial plate, which is the thing to point at if the room cannot read the panel from row four. Nobody watched it happen — the direction is the same in every series and it is still an inference.
how far back
eighteen months back · the day of surgery
Step back three · the operation itself ·
two minutes after the tooth came out
The implant went where the root had been. The root was
against the labial plate.
C · Monje 2025An implant placed in ungrafted narrow alveolar bone, with the
buccal surface now out in the mouth — a different published
case, reproduced because it is the clearest picture of this mechanism
in the corpus. Monje figure 2E(a). Monje et al. 2025. Reproduced under CC BY.
In the anterior maxilla the labial plate is often under a
millimetre before you start.
Observed an implant in a fresh socket, a buccal gap, and no graft in it. Inferred that grafting it would have held the contour.
Under 2 mm of buccal bone and part of the plate goes; under
1.5 mm, about 4 mm of it went. There was never 2 mm of it
to keep.
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+10:30. Anti-blame rung. The timing taxonomy is the NEXT slide, on its own, so keep this one on the photograph and the two millimetres. Resist the room's wish to convict here. Ask instead, by show of hands, who records the gap dimension in the operation note. THE QUOTE THE NUMBERS COME FROM: 'Buccal bone thickness should be at least 2 mm. If this thickness of bone is not available, some part of the buccal plate will be lost after remodeling, leading to a high risk of soft tissue recession' (Sangkhamanee 2024); where buccal bone was under 1.5 mm thick, about 4 mm of buccal bone was lost (Monje, in Monje 2025); and most peri-implantitis defects are a 2- or 3-wall configuration, the missing wall usually the buccal one (Monje 2025). NOT ON THE SLIDE, and it is the honest half of the amber line: Gaddale found NO significant difference in bone level change among the studies that DID fill the gap — so this rung is a real decision with a genuinely unsettled answer, not a mistake. ⚠ THIS IS WHERE THE CUT RUNG LIVES, AND IT IS NOT OPTIONAL. There used to be one more rung after this one — 'the day the tooth came out' — and its whole job was to stop this walk-back turning into an anti-immediate-placement lecture, which is exactly what a registrar will take away if you let them. SAY IT ALOUD, here, before you move to the principle: nobody made an error by placing on the day. Somebody saved her an operation and it was a defensible thing to do — one appointment instead of two, no second flap, no second consent, no healing phase, a tooth in the socket the same afternoon. And the evidence supports most of that: 16 studies, no crestal bone difference and no papillary difference against delayed placement, and about 1 mm LESS facial recession than a socket graft; papillae were more stable, or receded less, with a flapless approach than with a full-thickness flap (Gaddale 2024, SR and meta-analysis, aesthetic zone — closed access, numbers quoted, nothing reproduced). Immediate placement into infected sockets shows no significant survival difference, pooled survival 98% across 23 studies (Kaur 2021). What it does not buy is a quiet eighteen months: implant-related complications occurred MORE often with immediate placement plus provisionalisation, at the same survival. Surgical stage — immediate, early or delayed — did not yield a significant impact overall on peri-implant disease (Monje 2025), but one included study found immediate implants more likely to be diagnosed with peri-implantitis than delayed ones, OR 2.61, 95% CI 1.4 to 5.1 (Ortiz-Echeverri, in Monje 2025). Two systematic reviews of the same question that do not quite agree — that is what an unsettled decision looks like from the inside. AND THE COUNTERFACTUAL IS NOT AVAILABLE: nothing shows that delaying would have saved THIS site. The plate here was already thin, and a Type 2 placement into the same ridge inherits the same problem eight weeks later. The junction on the next slide is the teaching point, not the counterfactual. Then ask what should have been written in the notes on the day, and accept only concrete answers.
how far back
eighteen months back · the day of surgery
15 roads out of this junction. 3 taken.
When the implant went in, and what was done with the socket
Taken — 3 of 15
Immediate — Type 1, the day of extractionThe buccal gap left to fillImmediate provisional, non-occlusal
Available, not taken
Early — Type 2, four to eight weeksEarly — Type 3, twelve to sixteen weeksLate — Type 4, healed ridgeSpontaneous healing, then delayed placementSocket / alveolar ridge preservationAtraumatic extraction — periotome, piezoThe buccal gap graftedSimultaneous GBR at placementStaged GBR, then placeFlaplessFull-thickness flapSubmerged healing, second-stage uncovering
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
Thirty seconds. Point at the three lit chips first — immediate placement, the gap left to fill, an immediate non-occlusal provisional — and say that all three were taken in the same forty minutes. EVERY CHIP ON THIS BOARD WAS AVAILABLE ON THE DAY: that is the whole line, and it is what stops the rung before it becoming a conviction. This junction is doing the coverage work for the timing taxonomy for the entire hour, so it is worth its thirty seconds — but do not read the column out, point at the two or three the room will argue about and move.
how far back
eighteen months back · the day of surgery
The ballot, answered · walk-back 5 of 9
You voted before you were told anything. The walk ended at B.
AThe tissue was always too thin here — she needed a graft at the start
BIt should never have gone in the day the tooth came outwhere the walk ended
CThis is peri-implantitis — biofilm has got under the seal
DThe implant is in the wrong place
The tissue was always too thin describes the outcome, not the cause. It is thin because there is nothing behind it — a labial plate missing to the apex, and a hole sitting over it. You cannot graft your way out of a position.
Observed thin mucosa, and no labial plate behind it. Inferred that thicker tissue would have held over nothing.
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+13:30. Thirty seconds, and it is the debt the hour owes them for voting. Read the lit option out and then read the room's own distribution back to them. THE THIN-TISSUE ANSWER WILL HAVE WON — say the crimson line once and do not soften it: it is a description of the mucosa on the day she walked in, and a description of the end state is not a cause. ⚠ AND SAY THE OTHER HALF, because the lit option reads like a conviction and it is not one: the walk ended at the day of surgery because that is where the DECISION was, not because placing on the day was an error. You have already said, one rung back, that somebody saved her an operation and the evidence supports most of it. Say it again here in one sentence, or half the room leaves as anti-immediate-placement converts, which is the one takeaway this walk-back must not produce. Then straight into the principle — no pause between them.
how far back
before the tooth was ever extracted
The floor · foundation 4 of 8 ·
the soft tissue seal
The seal is the only barrier. It is
only as thick as the bone under it.
Between the biofilm and the bone there is one structure,
and it is weaker than the one the tooth had: the fibres run parallel
to the implant instead of inserting into it.
Observed a seal held up by whatever bone is behind it. Inferred that tissue alone rebuilds it — you cannot graft your way out of a position.
A · Wen 2024The same patient as the first photograph, eight years on,
after a pedicled palatal connective tissue flap. A seal can be rebuilt.
It costs another operation. Wen et al. 2024. Reproduced under CC BY.
Six of eight lit, and the walk was eighteen months.
Four walk-backs to go.
1Osseointegration1 walk-back
2Primary stability1 walk-back
3Bone and graft healing1 walk-back
4The soft tissue sealprimary here
5Load transfer without a PDLnot yet
6Three-dimensional positionalso lights
7Biofilm and the peri-implant lesionnot yet
8Anatomy and its limits3 walk-backs
← backwardsThe principlePlaced the same dayThe gap left openSitting too buccalThe tissue thinsThe outcome
+14:30. Let the headline sit before you speak, then advance once and let the board fill under it. This is the examinable line of the walk-back and it should be said once, slowly, and not paraphrased. WE WALKED PAST A RUNG to get here — 'placed the same day' — and the note on the PREVIOUS slide is where you pay that debt; do not arrive at the principle having let the room conclude that immediates are the error. THE ANATOMY, if you add a sentence: around a tooth the fibres insert, around an implant they lie alongside, and the vascularity is poorer. Wen puts the rest of it plainly — the interface is weakly attached by epithelial hemidesmosomes and lacks periodontal ligament and cementum. So it is held up by whatever bone is behind it, which is why a soft tissue problem in the aesthetic zone is so often a bone problem you are looking at through gum. THE COUNTER-EXAMPLE, off the slide now but the numbers are quotable: trans-mucosal tissue-level implants, full arch, 24 months — 100% survival, mean bone loss 1.19 mm, probing depths 2 to 3 mm, no plaque at any site (Carossa 2022, CC BY 4.0, and note it is an EDENTULOUS MANDIBLE, not an aesthetic zone; that mismatch is part of the point, because the seal is easiest to keep where the tissue is thick and nobody is looking). A short distance from the prosthetic margin to the crestal bone may predispose to peri-implantitis, which argues for tissue-level design or a transmucosal abutment (Monje 2025). Give the seal room and keep the interface away from the bone and it holds; ask it to stand over nothing and it will not. WHAT CHANGES ON FRIDAY — say all three, they are not on the slide. ONE: before you place into a socket in the anterior maxilla, measure the labial plate on the CBCT and write the millimetres in the notes; under 1.5 mm changes the conversation, not the torque value. TWO: make the gap a named decision — graft it or do not, and record which and why. THREE: when somebody shows you recession at eighteen months, ask for the scan before you offer the graft. ONE LAST OBSERVED AND INFERRED, about the whole walk-back: observed is two published cases, four reviews and a consensus systematic review that puts implant position ahead of every other surgical factor; inferred is that those three actions would have changed THIS outcome — two case reports cannot show that. What they can show is where the decisions were, and that the one everybody wanted to talk about was the last one in the chain. ASK THE ROOM to predict which card lights next; after five walk-backs they should be able to. Close on the photograph and on agency — the seal is rebuildable, at the price of another operation and eight years of somebody's patience. THE EIGHT-YEAR NUMBERS, off the caption now: pink aesthetic score 9, 5 mm probing depth without bleeding (Wen 2024). They drill at 14:00 and must leave with something to do, not something to fear. Then hand over: three years, and it bleeds when she brushes. Same implant, same position, one foundation further in.
how far back
Definitions · foundation 6 of eight ·
it has been earned; now name it
Three-dimensional position
Three-dimensional positionThe implant’s mesiodistal, buccolingual and apicocoronal position, plus its angulation.
Prosthetically drivenPlan the intended tooth first, then work backwards to the implant position that can support it.
Angular deviationThe angle between the planned implant axis and the axis actually placed.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · three years after surgery
Walk-back 6 of 9 · UR1 ·
a composite patient — two published cases
She is 44. The implant was placed three years ago.
A · Leventis 2025Bleeding on probing, 6–7 mm pocketing, and haemorrhagic and
purulent discharge at an anterior maxillary implant. Leventis et al. 2025. Reproduced under CC BY.
She has come back because it bleeds when she brushes, and because
there is a taste. Non-smoker. She attends every hygienist appointment she is
given.
Observed 6–7 mm circumferentially, bleeding, and pus. Inferred nothing yet — not the bone level, not the restoration, not the position, not who placed it.
Non-smoker. Attends every appointment she is given. None of the usual
risk factors are here.
+0:00. Show the photograph before you say anything. Let it sit and count to five. Give the history in the order written and then STOP — resist every instinct to add the radiograph. The vote on the next slide is worthless if they have been told enough to be right. SAY ALOUD, because the amber line is now one line: what you are being given is the depths, the bleeding and the discharge, and that is ALL you get — the bone level, the restoration, the three-dimensional position and the identity of whoever placed it are all still unknown to you and will stay unknown until you have voted. The published case is Leventis 2025, an infrapositioned anterior maxillary implant; our patient is a composite and every image says so on the slide.
how far back
today · nothing revealed yet
Walk-back 6 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AHer plaque control failed around a difficult crown—
BIt was overloaded — too much force, too little implant—
CThe restorative margin was too deep to keep clean—
DIt was in the wrong place before it was ever restored—
Hold on to what you picked. Two of those four are the same answer one rung
apart — and in ten minutes you will find out which of them came
first.
+1:30. Everyone votes — chase the abstainers, the mechanism depends on commitment. Do NOT comment on the distribution beyond reading it out. The bars are hidden until you advance one step, so nobody anchors on the early votes. Poll id w6. If the phones cannot reach the laptop (NHS guest wifi very often isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect the margin and the position to split the room — that split IS the walk-back, and the two options are one rung apart. NOT ON THE SLIDE, and only if you are challenged on why they are made to guess with too little information: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT); and simulation-first residents ended with higher knowledge than video-first despite worse initial performance (Ahn 2025, RCT). RESTATE THE RULE, it is walk-back 1's and it is not on this slide: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable. An M&M answers the first, arrives at 'bad dentistry', and stops. This walk-back goes three rungs past that.
how far back
three years earlier · the day it was cemented
Step back one · the decision to cement ·
we have walked past a rung without stopping — the restoration is already
off and on the bench
“Cement causes peri-implantitis” is not what the evidence says.
A · Leventis 2025The retrieved restoration. The margin is dark, contaminated, and
sits well apical to the tissue level. Leventis et al. 2025. Reproduced under CC BY. A different published case
— the patient in this walk-back is a composite.
Screw-retained against cement-retained, marginal bone loss: no
significant difference, p = 0.51. Choosing to cement has not been shown
to cost bone.
Observed the margin 3.5 mm subgingival, black, tissue inflamed around it. Inferred that the cement did it — the 81% figure counted only implants that already had disease.
So the fault is not the cement. It is a margin nobody can reach
— and that was decided by where the platform ended up.
+4:00. THE PIVOT OF THE WHOLE WALK-BACK. This is where the room wants to relax into 'bad dentistry' — do not let them, and do not let them leave with 'cement bad' either, because that is a slogan the current AO/AAP synthesis does not support. SAY THE 81% FIGURE OUT LOUD AND KILL IT IN PUBLIC: '81% of peri-implant disease is caused by cement' — the denominator in that study was implants that already had disease, so it cannot yield a risk at all, and the World Workshop grades the evidence LIMITED. Several of them will have quoted it in a viva, and the 11:55 handover session still asserts it. THE NUMBERS, if challenged — they are off the slide and in the appendix: screw- versus cement-retained, marginal bone loss, no significant difference, 6 articles, p = 0.51 (Lin 2025, AO/AAP SR + MA); residual cement IS associated with peri-implant disease in narrative review (Saleh 2022) and documented histologically in a single case (Guarnieri 2020) — association and an n = 1, not causation. So: choosing to cement has not been shown to cost bone; leaving cement where nobody can retrieve it is a different claim. AND THE TWO NUMBERS THAT MATTER FOR THE NEXT RUNG, also Lin 2025: emergence angle < 30°, concave or straight, carries lower peri-implantitis risk; abutment height ≥ 2 mm gives less marginal bone loss than < 2 mm, 12 articles, p < 0.0001. BOTH ARE DECIDED BY HOW DEEP AND HOW ANGLED THE PLATFORM IS. The prosthodontist inherits them; they do not choose them — say that sentence, it is the join to the next slide. FROM THE SLIDE THAT IS NO LONGER HERE: the margin is 3.5 mm subgingival on the palatal and deeper still interproximally, and nobody has been able to clean it since the day it was seated — not her, not the hygienist, not the dentist who cemented it. Ask whoever voted for her plaque control to look at that margin and say honestly what she was supposed to do about it with a toothbrush.
how far back
three years earlier · the day it was designed
Step back two · the day the crown was designed
The restorative dentist did not choose to cement. They were cornered into it.
The implant is angled labiallyas placed, not as planned
The screw channel exits the labial faceof a UR1
It cannot be screw-retainednot without a hole in the front of her tooth
So it is cementedand the margin goes deep to hide the metal
A margin nobody can cleanthree years before anyone finds out
Every one of those steps is reasonable. Each person made the best
decision available given what they were handed.
Observed a chain of decisions, each defensible on its own. Inferred a villain — and looking for one is how a room wastes an M&M.
+7:00. Name the anti-blame rule explicitly here — registrars in an M&M frame default to criticising the operator and stop there, and the walk-back only works if they keep going past that. The question is not who got this wrong, it is at which point the outcome stopped being avoidable: keep walking back until the answer changes from 'still avoidable' to 'already decided'. That is why this rung is worth its minutes — there is no villain in the chain, and that is the finding, not a softening of it. THE RETENTION TAXONOMY IS THE NEXT SLIDE, so stay on the chain here; walk down it once, out loud, one box at a time. IF YOU HAVE THE TIME, the second half of the original lede: this is what makes the case worth forty minutes rather than four — every decision in the chain is defensible on its own, and the outcome is still catastrophic.
how far back
three years earlier · the day it was designed
7 roads out of this junction. 2 taken.
How the crown was retained, and where its margin went
Taken — 2 of 7
Cement-retainedMargin 3.5 mm subgingival
Available, not taken
Angulated screw channelConometric / friction fitMargin at the tissue levelMargin 1 mm subgingival
Thirty seconds. Point at the DASHED chip first — screw-retained — and say why it is dashed: not because it is unproven, but because it is the road the angulation had ALREADY CLOSED before the prosthodontist ever saw the case. That single chip is the whole walk-back in one word. Then the one worth naming out loud: the angulated screw channel. It is the only road on this board that may still have been open, and nothing in the record says anyone considered it. Finish on the two lit chips — cement-retained, and a margin 3.5 mm subgingival — and let the room see that the second follows from the first.
how far back
three years earlier · the day of surgery
Step back three · the day of surgery · ninety seconds
of it · the floor of this walk-back
Everything you have just watched was decided here.
B · Fan 2023Entry, apex and angular deviation — planned position against
placed. Fan et al. 2023. Reproduced under CC BY 4.0.
Freehand, in a head-to-head RCT: 7.97° mean angular
deviation. Dynamic navigation in the same trial: 4.02°. Static guided pools
at 3.5°.
Observed planned position against placed — entry, apex, and the angle between them. Inferred that a guide would have saved this one. Nobody randomised this implant.
Arithmetic, not evidence. 8° over a 10 mm crown moves the
emergence of the screw channel about 1.4 mm — cingulum or facing,
and 8° spends it.
+10:00. FLAG THE ARITHMETIC BOX AS ARITHMETIC, OUT LOUD. 8° over a 10 mm crown height moving the screw channel ~1.4 mm is trigonometry, not a finding — but it is the budget they are working with, and whether 1.4 mm lands on the cingulum or through the facing depends on the tooth. It is millimetres either way, and 8° spends them. THIS IS THE SLIDE THAT SETS UP STATION A: at 14:00 they place guided and freehand and measure their OWN entry, apex and angular deviation against exactly this diagram — say so here, by name, so the afternoon has somewhere to land. THE FULL TABLE, off the slide and in the appendix: 3.5° static guided, mean angular deviation (Tahmaseb 2018, 20 studies, 2,238 implants); 3.68° dynamic navigation (Jorba-García 2021, SR + MA); 4.02° dynamic and 7.97° freehand in the same head-to-head RCT (Jorba-García 2023). Guidance roughly halves angular error, and it costs FOURTEEN MINUTES of operating time with no improvement in patient satisfaction or post-operative pain — that is the honest price and it must be said. This is deliberately the same evidence table as fork 7 in the decision-session sample: if both sessions survive, the room meets these four numbers forwards and backwards. THE GUIDANCE FAN IS NO LONGER ON SCREEN — one fan per walk-back, and the screw-channel rung has no photograph while this one has the best diagram in the corpus. Its seven roads still print in the appendix, and its point is one sentence you should say anyway: freehand placement is the road that leaves NO RECORD OF WHAT WAS INTENDED, which is why nobody can now tell you whether this implant went where it was planned to go.
how far back
three years earlier · the day of surgery
The ballot, answered · walk-back 6 of 9
You voted before you were told anything. The walk ended at D.
AHer plaque control failed around a difficult crown
BIt was overloaded — too much force, too little implant
CThe restorative margin was too deep to keep clean
DIt was in the wrong place before it was ever restoredwhere the walk ended
Cement causes peri-implantitis is a slogan the evidence does not carry — screw against cement shows no difference in bone loss. And no toothbrush reaches a margin 3.5 mm subgingival: her plaque control was never the variable.
Observed a contaminated margin 3.5 mm subgingival, and a bleeding implant. Inferred that the cement did it, or that she did.
+12:00. Thirty seconds, and it is the debt the hour owes them for voting. Read the lit option out, then read the room's own distribution back to them. TWO WRONG ANSWERS CAME INTO THIS ROOM AND BOTH ARE ON THE BALLOT. The first is hers — a non-smoker who attends every appointment she is given, blamed for a margin she was never able to reach; ask again what she was supposed to do about it with a toothbrush. The second is the one they will carry to a viva, so kill it here for the second time and in public: 'cement causes peri-implantitis' is not what Lin 2025 says, and the 81% study could not have produced a risk at all. THE MARGIN IS STILL NOT THE FLOOR either — it was inherited from an angle chosen months earlier, which is what the lit option says and what the principle is about to generalise. Go straight into it; no pause.
how far back
three years earlier · before the drill went in
The floor · foundation 6 of 8 ·
three-dimensional position
Everything downstream is decided
before the drill goes in.
Of every surgical and implant-related factor examined across 34
studies, implant position is the one the evidence proves predisposes to
peri-implantitis.
Observed a consensus review naming position as the one proven surgical factor. Inferred that anything downstream fixes it — even mucositis resolution is “elusive”.
Seven of the eight are lit — and one card has never lit at all.
Three walk-backs to go.
+13:00. Let the headline sit before you speak, then advance once and let the board fill under it. This is the only slide in the walk-back with a claim as a headline — that is the format's payoff and it should feel like an arrival, not a summary. AND IT IS NOT A TEACHING DEVICE: Monje 2025, AO/AAP consensus systematic review, 34 articles — of all surgical and implant-related factors examined, inadequate implant position is the ONE that clinical evidence proves predisposes to peri-implantitis (implants in regenerated bone to a lesser extent; implant-related factors had insufficient evidence). That is the strongest support this claim has ever had. SAY WHY IT IS FOUNDATION 6 AND NOT A PROSTHODONTIC NICETY: position is the variable that sets the emergence angle, the abutment height, the retrievability of the margin and the cleansability of the whole unit — before any of those people have been asked their opinion. THE UNCOMFORTABLE HALF, which is the amber line in full: adjunctive treatment of peri-implant MUCOSITIS — the reversible one — gives no clinically significant benefit over debridement alone, and complete resolution is described in the current synthesis as 'an elusive outcome' (Lin 2025, AAP/AO SR + MA, 25 articles). Let them draw their own conclusion about peri-implantitis. WHAT CHANGES ON FRIDAY — say it, it is not on the slide, and it is where the agency comes back: plan the restoration before you plan the implant; if the screw channel will not exit through the cingulum, the implant is in the wrong place and no prosthetic ingenuity will fix it; and if you do end up cementing, the margin goes where you can reach it, even if the metal shows. ONE LAST TIME ABOUT THE WHOLE WALK-BACK: what was OBSERVED is a bleeding implant at three years, a contaminated margin on the bench, and a consensus review naming position; what is INFERRED is the chain between them — nobody photographed this implant going in, and no case report can show that a guide would have prevented it. What it shows is WHERE THE DECISIONS WERE, which is the thing they are here to be able to see. THE BOARD IS THE ARGUMENT, not a scorecard. This one implant broke position, and through it the seal, and through that the biofilm equilibrium — in that order. Six walk-backs in, the room can predict which cards light next: ASK THEM BEFORE YOU ADVANCE, and the one card still dark is the next walk-back's primary. Then hand over: three walk-backs follow, each reaching further back than the last — a screw that snapped, a patient nobody ever saw again, and a mandible that broke before she sat down. Different photographs. Watch how few principles they need between them.
how far back
Definitions · foundation 7 of eight ·
it has been earned; now name it
Biofilm and the peri-implant lesion
Peri-implant mucositisBiofilm-associated inflammation confined to the soft tissues, without continuing supporting-bone loss.
Peri-implantitisBiofilm-associated inflammation around an implant with progressive loss of supporting bone.
Baseline radiographThe reference image after restoration or loading against which later bone levels are compared.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · four years after it was fitted
Walk-back 7 of 9 · the maxilla, under an overdenture
· a composite patient — one published case
She is 67, and the denture has been coming loose for four years.
A · Alwaqyan 2026“…the fractured prosthetic screw fragment lodged
within the internal connection… The implant fixture remained
osseointegrated with maintained surrounding peri-implant bone support”
— the paper’s own caption, cropped to the fixture. The words
elided from it name a site we come back to. Alwaqyan et al. 2026. Reproduced under CC BY 4.0.
Diabetic, a heavy smoker. Four years ago the upper arch was
restored as an overdenture clipped onto implants. Relined, adjusted, inserts
replaced, tissue lasered — and this week it would not seat.
Observed a fixture still integrated, and a fractured screw fragment inside its connection. Inferred everything else — the implant count, the occlusion, how long it was loose.
Four years of adjustments, and the fixture is still osseointegrated.
The bone did its job. The hardware did not.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+0:00. Put the film up in silence and count to five. Somebody will say “that’s just an implant” — good, say nothing, because that IS the finding: the fracture is inside the connection and a periapical barely shows it, which is part of why nobody catches these early. Give the history in the order written and then STOP. Every extra fact you add makes the vote on the next slide worth less. Do not use the word overload yet. NOT ON THE SLIDE: her lower arch was restored at the same time, and the gum around the attachment was trimmed with a laser more than once. THAT IS ALL YOU GET, and say it in these words if the room starts guessing — what is observed is a fixture that is still integrated, with bone up to its shoulder, and an author who reads a fractured screw fragment inside its connection; what is NOT YET KNOWN is how many implants are under that prosthesis, what her occlusion was, how long the screw had been loose, or whether it had ever been changed.
how far back
today · nothing revealed yet
Walk-back 7 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
ATwo implants were carrying a whole upper arch — the load had nowhere else to go—
BIt was never torqued properly when the prosthesis was made—
CIt had been loosening for years and nobody asked why—
DHer occlusion was traumatic and nobody adjusted it—
Hold on to what you picked. Two of those four are the same answer one
step apart — and one of them is the answer the literature cannot
settle.
Scan once — it follows the deck
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+1:30. Chase abstainers — commitment is the whole mechanism. The bars stay hidden until you advance one step, so nobody anchors on the early votes. Read the distribution out and say nothing about it. Poll id w7. If the phones cannot reach the laptop (NHS guest wifi isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect a heavy vote for B and D. B is the blame answer and it is unfalsifiable; D is the answer the field itself is still arguing about, and we spend a whole slide on why that is worth knowing. SAY THE RULE ALOUD, it is no longer on the slide: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — those are different questions and only one of them is answerable. This is the walk-back where the M&M answer is most tempting, because there is a torque figure and somebody did or did not hit it. THE EVIDENCE FOR VOTING FIRST, if challenged: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm. That is why you vote first, every time, and why being wrong is the mechanism rather than an embarrassment.
how far back
four years of appointments · none of them about the screw
Step back one · the four years before this morning
· the appointments nobody counted
Every one of those visits fixed something.
Not one of them fixed the screw.
A · Alwaqyan 2026This morning, at the end of those four years: the fractured screw
fragment, lifted out of the connection. It came out whole and the connection
was undamaged — in one published case, by one operator. Alwaqyan et al. 2026. Reproduced under CC BY 4.0.
Five years in function: relines, attachment adjustments, locator
inserts, laser trimming — and, verbatim, “no documented previous
replacement of the fractured prosthetic screw.”
Observed the maintenance record, and that one sentence in it. Inferred that the screw fatigued — the author infers it too, and nobody measured a preload.
A screw that comes loose once is a repair. A screw that comes loose twice
is a reading — of the occlusion, the fit, or how many implants are
under it.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+3:30. This is the slide the walk-back exists for. Hold the taxonomy — the junction that follows names every road out of this appointment, and it is where you spend the thirty seconds. Here, just let the room hear that everything done was a repair and everything not done was a question. Ask by show of hands who has retightened the same screw twice without taking a radiograph. Expect most of the room. Do not let this become a criticism of general practice; the appointment is fifteen minutes long and the patient came in for a loose denture. THE FULL RECORD, verbatim, since only half of it fits on the slide: a prosthesis in function about five years with documented relining, attachment adjustment, locator insert replacement, recurrent loss of retention, soft tissue overgrowth and laser management of it — and “there was no documented previous replacement of the fractured prosthetic screw since the fabrication of the prosthesis.” The author's own inference is that repeated insertion and removal, long service and posterior functional loading were “considered contributing factors”. THE NUMBERS, if challenged: abutment screw loosening 16.2% across 18 studies, one of the commonest mechanical complications, and mechanical complications cluster in the first nine years (Verma 2023, SR); and across 33 studies of overdenture maintenance “a mean complication rate was impossible to determine” and “no clear identification of the causes of mechanical complications was found” (Assaf 2017, SR). So the loosening was ordinary, and the field cannot tell you what causes it. Both of those are reasons to write it down, not reasons to stop looking. WHAT THE PHOTOGRAPH COST, which used to be a slide of its own — ninety seconds, and its job was to price the complication, not to teach the technique. Soft tissue was cleared with a laser to expose the connection; the fragment was troughed around and turned counterclockwise with an ultrasonic tip, engaged with cyanoacrylate on a flexible micro-applicator, and lifted out. THE REGISTRARS WILL WANT THAT LIST AND YOU SHOULD REFUSE TO GIVE IT, because the same paper lists what the same instruments do when it goes the other way: damage to the internal connection, compromised restorability, limited tactile control, overheating, or inadvertent apical displacement of the fragment — and then the implant comes out. Read the good outcome carefully: the fragment was retrieved, the connection was intact with no visible deformation, and a new screw and attachment went back in at the manufacturer's torque. What is NOT shown is that any of it is reproducible. One operator, one connection design, one fragment that happened to have mobilised. A technique that works once is a technique, not a rate, and there is no denominator anywhere in this project for how often retrieval fails. If anyone asks whether they should attempt this in their own clinic, the honest answer is that the paper is a single case report.
how far back
four years of appointments · none of them about the screw
13 roads out of this junction. 4 taken.
A screw that keeps coming loose
Taken — 4 of 13
Replace the locator insertAdjust the attachmentReline the dentureTrim the overgrown tissue
Available, not taken
Retighten it to torqueReplace the screwRadiograph it and record the patternCheck and adjust the occlusionVerify passive fitAsk about parafunctionAdd an implant or change the designNightguardSend it back to whoever planned it
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+5:00. Thirty seconds, no more. Point at the lit chips first — the 4 roads that were taken, all four of them repairs — then at the count above them. Every road taken was a repair. Not one of them was a question. Then move your hand across to the unlit column and read the 9 that were not taken, because that column is the argument: Retighten it to torque · Replace the screw · Radiograph it and record the pattern · Check and adjust the occlusion · Verify passive fit · Ask about parafunction · Add an implant or change the design · Nightguard · Send it back to whoever planned it. Do not let this become a criticism of general practice; every one of those appointments was fifteen minutes long and the patient came in for a loose denture. The two roads at the end of the unlit column — nightguard, and send it back to whoever planned it — are the ones registrars forget they are allowed to take.
how far back
four years earlier · the day it was designed
Step back two · four years earlier · what it sat on,
and what was built on top of it
Everything at this junction was chosen well.
None of it was ever about screws.
Internal conical connections lost less marginal bone than
external, 20 articles. Non-splinted restorations lost less than splinted,
8 articles. Crown-to-implant ratio: no significant effect.
Observed a bone-level tapered fixture, 4.1 × 10 mm, internal conical, torqued to the manufacturer’s figure. Inferred that a bar, or splinting, would have saved the screw.
Every one of those findings is a measurement of marginal bone level.
Not one of them was ever asked to predict a fracture.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+7:00. TWO RUNGS IN ONE — the hardware and the design were saying the same sentence. The connection taxonomy is the junction slide after this one; here, stay on the numbers. SAY THE SPLINTING FINDING TWICE. The room will not believe it the first time, and a registrar who has been taught to splint everything posterior needs to hear that the meta-analysis went the other way on bone loss. Then hand the nuance straight back: it is a bone-level outcome, not a mechanical one, and this walk-back is about a mechanical failure. Splint it because the mechanics ask you to, if you like. Do not splint it because you think it protects bone. THE NUMBERS, if challenged, all Lin 2025 unless said otherwise: internal — particularly conical — connections lost less marginal bone than external, 20 articles, p < 0.0001; platform switching, 20 articles; abutment height ≥2 mm, 12 articles; one abutment, one time rather than repeated disconnection, 10 articles, all p < 0.0001; emergence angle under 30° with a concave or straight profile, lower peri-implantitis risk; non-splinted showed less marginal bone loss than splinted, 8 articles, p = 0.04; crown-to-implant ratio no significant effect, 5 articles, p = 0.32. And from Verma 2023, SR: mechanical failure of implant-supported overdentures runs about 3% across 18 studies, against an overall mechanical failure prevalence of 5.6 to 7.7%. TWO THINGS THE CHIPS HIDE. First, the tree marks platform switching as thin — that was a statement about how many papers this corpus holds, not about the field, and round 1 corrected it in writing. Twenty articles is not thin. “Thin in my reading” and “thin in the literature” are different claims and it is worth hearing one mistaken for the other. Second, Lin 2025 is free to read and NOT free to re-use: its licence permits neither reproduction nor adaptation of a figure, so the numbers are quotable and the figures are not. That is why there is no forest plot and no photograph on this slide. THE NUMBER THIS PROJECT DOES NOT HAVE: there is no verified figure anywhere in this day for how long a cantilever may be. The map marks cantilever length limits as case-series evidence and no limit was ever sourced. If someone hands you a multiple of the anterior-posterior spread, ask them for the paper. THE DESIGN FAN, which is no longer on the screen — the road taken was an overdenture on two implants, and the roads that existed were: Single crown · Implant-supported FPD · Cantilever FPD · Full-arch fixed — metal-acrylic · Full-arch fixed — monolithic zirconia · Overdenture on four implants or a bar · Conventional complete denture · Cantilever length limits · Cross-arch stabilisation · Occlusal scheme — mutually protected or group function · Cusp inclination and table width · Implant-protected occlusion · Parafunction management and nightguard · Occlusal adjustment at review · Progressive loading. Two of those are dashed on the map because the field talks about them and the evidence does not put them there: implant-protected occlusion and progressive loading. SAY ALOUD: the only occlusal assessment recorded in the whole history was made on the day the screw was found broken. It found no gross discrepancy and no premature contact — four years too late to be evidence of anything.
how far back
four years earlier · the day it was designed
11 roads out of this junction. 4 taken.
What the crown sat on
Taken — 4 of 11
Conical / Morse taperBone-levelTwo-pieceScrew material, torque and preload
Available, not taken
Internal hexPlatform switchingMicrogap and micromovementTissue-levelOne-pieceAnti-rotational indexing
On the map, not in the evidence
External hex
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+9:00. Twenty-five seconds. Point at the lit chips first: the paper names all of it — bone-level, conical, two-piece, and a screw with a material, a torque and a preload. Then say why this slide exists: this is the one place in the whole hour that connection geometry can live, and it is here on purpose (15-coverage-audit.md §4). The dashed chip is external hex — a road the field has largely left, not a road she was denied, and say that distinction out loud or somebody will write it down as a criticism. Do not read the middle column; sweep your hand across it and let the room see how many choices were made without anybody once asking what would happen to the screw.
how far back
four years earlier · before anything was placed
Step back three · before a single implant was ordered
· the decision that fixed everything downstream
Two above. Four below.
A · Alwaqyan 2026“Panoramic radiographic evaluation demonstrating maxillary
and mandibular implant-supported prosthetic rehabilitation” — the
paper’s own caption. Count them before you are told a number. Alwaqyan et al. 2026. Reproduced under CC BY 4.0.
Both maxillary implants sit well forward of a pneumatised
antrum. Everything behind them is denture on mucosa.
The prosthesis extends well past bothso a posterior contact is a lever about the attachment
The moment ends at the terminal screwand nothing between it and the bone takes it
Observed two implants above and four below, on this film. Inferred the paper’s own text — it puts the screw at UR7, with no implant near it.
This is where the walk-back leaves prosthetics and arrives at
foundation 6 — settled before anyone chose an attachment.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+10:30. Put the panoramic up and ask the room to COUNT before you say a number. They will count four in the mandible first. Then read the amber line. The discrepancy between the paper's text and its own figure is not a gotcha — it is the whole observed-versus-inferred rule made concrete on a published paper, and it is the most valuable thirty seconds in this walk-back. The text places the fractured screw at “the maxillary right second molar implant region” — that is UR7; its own panoramic shows no implant anywhere near UR7 — two maxillary implants, both well forward, under an antrum that fills the whole posterior maxilla. We cannot resolve that from outside the case. What is not in doubt is two above and four below, on this film. SAY IT IN THESE WORDS: published case reports are written backwards too, by the person it happened to, and the rule applies to them exactly as it applies to us. Do not turn it into a criticism of the author; a single-author case report has no second reader. NOT ON THE SLIDE, the first link of the chain: two implants, both forward of the sinus, means everything behind them is denture on mucosa — which is why the load path ends where it does. Where the implants went, and how many there were, was settled before anyone chose an attachment, and it set the load path for every year that followed.
how far back
the literature · 2019 against 2025
Same rung · the literature · and it does not agree
with itself
Two systematic reviews, six years apart, on how much load is too much.
Bertolini 2019clinical studies that met the criteria2
Bertolini 2019animal studies that met the criteria4
Mojaver 2025studies included — six years later80
Studies admitted by each review’s inclusion criteria.
The criteria make the field. Bertolini found 6 in total; Mojaver, 80.
Observed two reviews, both real, both careful, reaching different confidence. Inferred every number anyone gives you for how much load is too much.
Two studies against eighty. The inclusion criteria make the field —
and what each review then concluded is on the next slide.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+13:30. The most intellectually useful slide in the walk-back and the easiest to rush. GIVE IT TWO MINUTES. Ask the room which review they would cite in a medico-legal report and then ask why. They are not asking an identical question — and that is the point: the inclusion criteria make the field. THE HONEST POSITION TO MODEL, and say it in these words: you adjust occlusion on implants because the mechanics make sense and the intervention is cheap, NOT because it is proven — and saying so does not weaken you in front of a registrar, it teaches them how to read. SAY THIS PART OUT LOUD, it is no longer on the slide: implant-protected occlusion is near-universal doctrine on thin evidence. That is this project's own note against the concept, and neither review above supports the confidence with which it is taught. THE CONSEQUENCE LINE, also off the slide: the force has to go somewhere and there is no ligament to put it in — but if you are told a number for how much is too much, ask where it came from. This is what a field looks like while it is still arguing, and registrars almost never get shown one. Mojaver 2025 is doi 10.1111/jopr.14088 if anyone wants it.
how far back
the literature · 2019 against 2025
Same rung · the two reviews, in their own words
Same question, six years apart, opposite confidence.
little
The human effect is
“poorly reported and provides little evidence to support a
cause-and-effect relationship” — though animal work showed
bone loss with excessive overload even without inflammation
Bertolini 2019
0.65–1.20
mm of marginal bone loss attributed to
occlusal factors; 1.0 to 3.0 mm with traumatic occlusal forces,
peri-implantitis 20–50%, and a described synergy between mechanical
overload and biofilm
Mojaver 2025
Observed two reviews, both real, both careful. Inferred every number anyone gives you for how much load is too much.
Neither supports the confidence with which implant-protected occlusion is
taught. If you are given a number for how much is too much, ask where it
came from.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+15:00. THIS is the two minutes, not the slide before it. Ask the room which review they would cite in a medico-legal report, and then ask why. They are not asking an identical question, and that is the point. THE HONEST POSITION TO MODEL, in these words: you adjust occlusion on implants because the mechanics make sense and the intervention is cheap, NOT because it is proven — and saying so does not weaken you in front of a registrar, it teaches them how to read. This is what a field looks like while it is still arguing, and they almost never get shown one. Mojaver 2025 is doi 10.1111/jopr.14088.
how far back
four years earlier · before anything was placed
The ballot, answered · walk-back 7 of 9
You voted before you were told anything. The walk ended at A.
ATwo implants were carrying a whole upper arch — the load had nowhere else to gowhere the walk ended
BIt was never torqued properly when the prosthesis was made
CIt had been loosening for years and nobody asked why
DHer occlusion was traumatic and nobody adjusted it
Torque and occlusion are the two answers this room reaches for, and neither survives the record: nobody measured a preload, and the only occlusal assessment in the whole history was made on the day the screw was found broken. Both sit downstream of a load path that was fixed the day two implants were chosen for a whole arch.
Observed a fractured screw, no preload ever measured, and one occlusal note made that day. Inferred that it was under-torqued, or that her occlusion did it.
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+16:00. Fifteen seconds, and do not teach off it — the walk has already made the argument, this only closes the loop the ballot opened. Read the lit row out and stop. If the blame answer took the room, name that out loud: it is unfalsifiable four years later, and being unfalsifiable is exactly why it feels safe to give. Anyone who voted for the occlusion was not wrong about the mechanism, only about the depth — and the slide before this one is why we cannot put a number on it either. Then go straight to the principle.
how far back
four years earlier · before any of it
The floor · foundation 5 of 8 ·
load transfer without a PDL
There is no ligament. The load ends in bone, or it ends in metal.
A tooth’s ligament moves, intrudes, absorbs and reports
back. An implant has none of it. Force is damped nowhere: it ends in bone, or in
the weakest metal in the load path.
Observed bone to the shoulder after four years. The screw failed, the interface did not. Inferred that this was a disaster.
A component reported a load problem for four years in the only language it has
— coming loose. Eight of eight lit, with two walk-backs still to come.
1Osseointegration1 walk-back
2Primary stability1 walk-back
3Bone and graft healing1 walk-back
4The soft tissue seal3 walk-backs
5Load transfer without a PDLprimary here
6Three-dimensional positionalso lights
7Biofilm and the peri-implant lesion1 walk-back
8Anatomy and its limits3 walk-backs
← backwardsThe principleHow many implantsThe prosthesisWhat it sat onThe looseningThe outcome
+16:30. Let the headline sit before you speak, then advance once and let the board fill under it. THE BOARD FILLS HERE — every one of the eight foundations has now been reached, on walk-back seven of nine, and the two that remain will light cards that are already lit. Say that out loud: the argument of the hour is not that there are lots of principles, it is that there are eight and nine complications keep landing on the same ones. FINISH THE CRIMSON LINE — the slide stops at “coming loose”, and the end of the sentence is: and every appointment answered the symptom instead of the sentence. IF YOU SAY ONE MORE SENTENCE than is on the slide, say this: a loose screw is the only warning this system is capable of giving you, and it is the one warning we are trained to silence in under a minute. THE GOOD ENDING, which used to be a photograph on the old convergence slide and is now a sentence: the postoperative film shows the fixture intact, bone maintained, no fracture and no loss of integration, and a new screw and attachment back in at the manufacturer's torque. THE REST OF THE PRINCIPLE, which is no longer on the slide: no proprioception — osseoperception instead, with a tactile threshold roughly an order of magnitude worse; no orthodontic movement, no intrusion, no shock absorption. Here the weakest piece of metal was a prosthetic screw, the one component in the assembly that is MEANT to be replaced, which is the only reason this walk-back has a good ending. THE EVIDENCE BEHIND THE BOARD: screw loosening 16.2% and mechanical complications clustering in the first nine years (Verma 2023); how much occlusal load causes bone loss in humans is contested — two systematic reviews, two answers (Bertolini 2019, Mojaver 2025). The principle is not contested. The threshold is. Teach the first with confidence and the second with the argument attached. WHAT CHANGES ON FRIDAY, three things, none of them “be careful” — ONE: write the maintenance history on one line in the notes, date, what you adjusted, and what you did not check; three retention failures in a year is a diagnosis waiting to be made by whoever reads that line. TWO: the second time a screw loosens, take a radiograph and check the occlusion BEFORE you reach for the driver. THREE: at planning, write down the load path — how many implants, what extends past the last one, and what is in the opposing arch; if you cannot write it, you have not planned it. AND THE HONEST CAVEAT: this is one published case, one implant, one retrieved fragment, and a literature that agrees screw loosening is common and disagrees about what causes it. That those three actions would have prevented this is inferred; a case report cannot show it and neither can nine of them. What they can show is where the decisions were. Then hand over: eight years, and nobody ever saw her again — watch what happens when the walk gets long enough that the surgeon is no longer in it.
how far back
Definitions · foundation 5 of eight ·
it has been earned; now name it
Load transfer without a PDL
Periodontal ligament · PDLThe fibrous ligament that suspends a tooth in its socket. An osseointegrated implant has none.
PreloadThe tensile force created in a screw when it is torqued, clamping the components together.
CantileverPart of a prosthesis extending beyond its last supporting implant, creating a lever.
DEFINITION BEAT. Fifteen seconds; twenty-five at the outside. Read only the first term and its definition. Let the room read the other two, then advance. Do not add examples and do not reopen the case — the walk-back has already done the teaching. This slide exists only so the vocabulary is explicit.
how far back
today · eight years after placement
Walk-back 8 of 9 · LR6 and LR7 ·
a composite patient — two published cases
You placed two implants eight years ago.
You last saw her at one year.
A · Fernandes 2025The periapicalplatform to bone, in mm
Figure 7; the measurements are the figure’s own. Fernandes et al. 2025. Reproduced under CC BY.
She is 71. She came back three weeks ago because of the
smell. Bleeding and suppuration at both. Platform to bone,
7.8 and 9.2 mm — on 13 mm implants, that is more than
60% and more than 70% of their length.
Observed bleeding, suppuration, and bone gone at both. Inferred nothing yet — not her risk, and not whose job she was.
Two implants, eight years, one review. Seven of those years are simply
missing from the record.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+0:00. Put the radiograph up in silence and count to five before you speak. Give the history in the order it is written and then STOP — every extra fact makes the vote worth less. Resist the urge to say the word 'maintenance' anywhere on this slide. By now the room expects a surgical answer, because seven walk-backs have given them one. Let them expect it. NOT ON THE SLIDE, and each worth one sentence: the bridge is one piece and there is nowhere to put a brush — that is foundation 4 lighting again, the seal she needed was the one thing the prosthesis design made impossible to clean; and she is not in pain and has not been in pain, which is why she did not come earlier. THE FULL AMBER LINE, if you want to read it out: observed — bleeding, suppuration, and 7.8 and 9.2 mm of bone gone from the platform at two 13 mm implants; not yet known — where her bone level was at one year, what her periodontal history was, whether she smoked, or whose job it was to see her.
how far back
today · nothing revealed yet
Walk-back 8 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
AThe implants were badly placed and the bridge could never be cleaned—
BShe had peri-implantitis for years and it was caught far too late—
CHer risk was never assessed, so there was nothing to set an interval from—
DShe was discharged at one year with no recall interval—
Three of those four are the same answer at three different depths. The fourth
is the one this room is trained to give — and we walk straight
past it.
Scan once — it follows the deck
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+1:30. Chase abstainers. The bars stay hidden until you advance one step, so nobody anchors. Read the distribution out and say nothing about it. Poll id w8. If the phones cannot reach the laptop, press H and type counts with 1–9; 0 clears, V opens or closes voting. Expect a heavy vote for A. A is the trap, and it is the whole reason this walk-back exists: it is not the answer, and refusing to let it be the answer is the point of the next twenty minutes. TWO THINGS THAT ARE NO LONGER ON THE SLIDE. First, the rule for this vote, and say it in these words: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — those are different questions and only one of them is answerable. Second, name the expectation before you break it: seven walk-backs have ended somewhere you could have stood with an instrument in your hand; hold that expectation, because it is about to be useless. The generation evidence, if anyone asks why they vote before being taught: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). Watching is the losing arm.
how far back
the eight years, compressed into one slide
Straight after the vote · the only slide here that is not a
step backwards · foundation 7 · concept 7.2
It is not periodontitis with a screw in it.
B · Park 2023The film that showed
nothingat prosthesis deliveryB · Park 2023Twenty-six months
lateran ovoid lucency around the body
A different published case and an unusual pattern — a
lateral lesion with the crest still intact. The pair is the point, not the shape:
a film with nothing on it, a film with everything on it, and nothing in between.
Panels cropped from figure 5. Park et al. 2023. Reproduced under CC BY 4.0.
More than twice as large as a periodontal lesion, hotter, and it
moves in steps rather than down a slope.
Observed larger, and faster — both sourced in this project. Inferred closer to the crest, poorly encapsulated — described, not sourced here.
A recall interval that is safe for a periodontal patient is not automatically
safe here. Eight quiet years is not eight years of nothing happening.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+3:00. The extra slide, and it is not optional — without it this walk-back asserts a behaviour it never explains. Sixty seconds. PUT THE TWO RADIOGRAPHS UP FIRST AND ASK THE ROOM WHICH ONE IS ABNORMAL BEFORE YOU SAY ANYTHING; about half will hesitate on the second. Park's own words about that pair: “the actual bone resorption process seems to have started to occur earlier than this.” THE EVIDENCE, which is no longer a table on the slide, in the order to say it: “peri-implantitis lesions are more than twice as large as those seen with periodontitis” (Monje 2025, AO/AAP SR); more plasma cells, macrophages and neutrophils, and more highly vascularised lateral to the cellular infiltrate than periodontitis sites (Monje 2025); it “progresses faster than periodontitis, being more frequently found on a hastened, unsteady pattern” (Fernandes 2025); and experimental mucositis is only partially reversible and takes twice as long to resolve as experimental gingivitis (Monje 2025). Bigger, hotter, faster, and it moves in steps rather than down a slope. THE HONESTY IN THE AMBER LINE: the foundations tree gives four differences. Two are sourced in this project — the lesion is larger, and it progresses faster and less predictably. Two are descriptive and not separately sourced here — that it extends closer to the crest, and that it is poorly encapsulated. Both are the standard reading of the same biopsy literature; neither has a citation in this hour, so do not put a number on them. Do NOT let anyone leave thinking peri-implantitis is periodontitis around titanium — that mistake is what makes a twelve-month recall feel adequate.
how far back
three weeks ago · the honest prognosis
Step back · three weeks ago ·
before you promise her anything
Treating the failing implant is better evidenced than preventing
the failure.
A · Fernandes 2025Twelve months after implantoplasty in a different case in the
same series: no bleeding, no suppuration, probing depths of 1 to 2 mm
— and the implant bodies are now in her mouth. This is what a good outcome
looks like. Fernandes et al. 2025. Reproduced under CC BY.
Implantoplasty with regeneration against without: survival
97% and 94%. Adjunctive treatment of mucositis — the
reversible one — gives no clinically significant benefit.
Observed a 97% figure for the salvage operation. Inferred that it applies to her — nobody randomised this patient.
Whichever road is taken, the ridge she leaves with is shorter, thinner and
scarred against the one she arrived with. There is no road back.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+5:30. THE PIVOT. Set the scene in one line before the slide does anything — she is sitting in your chair and something has to be decided. The eighteen roads out of this junction are the slide after this one; here, hold the room on the two numbers and the photograph. The room will hear 97% and relax; do not let them. Read the mucositis clause immediately after the survival figure, out loud, and leave the pause: if we cannot reliably resolve the reversible one, draw your own conclusion about the other. Lin 2025 (AAP/AO SR+MA, 25 articles) also records that complete resolution of peri-implant mucositis is “an elusive outcome” — that is the phrase to say, and it is the honest end of the same literature that produced the 97%. Then say it plainly: we are better at rescuing the failing implant than at stopping it failing. That is uncomfortable, it is true, and it is the reason the rest of this walk-back is about appointments rather than instruments. THE PHOTOGRAPH IS DOING THE WORK — that is a success, and it still leaves her with bare metal in her mouth for the rest of her life. On the network map the passenger rejoins the Bone line one zone worse; that is concept 7.5, and it is the last thing this hour has to teach about failure. If one thing survives from this walk-back it should be that there is no road back to the ridge she had. TWO MORE NUMBERS, off the slide, if you are challenged. Fernandes figure 14: 7.8 and 9.2 mm of bone gone from 13 mm implants — 60% and 71% of the implant — treated by implantoplasty and graft, 0.0 and 1.7 mm at 14 months, still holding at 43 months. And Hakkers 2025, a case series of three reconstructive cases: pockets fell by 4 to 6 mm and suppuration stopped, and none of the three met the guideline definition of a treated site at one year, with 1.0 to 4.0 mm of recession. Someone usually objects that a series of three is weak; agree, and point out that it is the honest end of the same literature.
+7:30. Forty seconds — the biggest junction in the deck and the last one of the hour, so it is worth more than the usual thirty. Point at the two lit chips first: non-surgical debridement, and what you tell the patient. Then run your hand down the middle column so the room sees how many roads exist, and stop deliberately on the four dashed ones. WHY THEY ARE DASHED: implantoplasty is sourced; explantation and re-implantation are not. Round 2 of this project's evidence check closes the implantoplasty question and records removal and re-implantation as still open — the corpus holds two papers that touch explantation and none that give an outcome. So say that in the room: we do not have a good number for what happens after you take it out. That is a legitimate answer and it is more useful than a confident one. THEN GO BACK TO THE SENTENCE YOU HAVE JUST READ on the rung before this one, and say it over the top of the whole board: whichever of these roads is taken, the ridge she leaves with is shorter, thinner and scarred. There is no road back, and there is no chip on this slide that puts one there.
how far back
seven years earlier · the one-year review
Step back four · one year after the bridge went in ·
nothing here was negligent
“Doing well. Discharged to your care for routine review.”
Implants that lost ≥0.5 mm in the first year of
function had five times the odds of peri-implantitis over the following
ten years. She had that film taken. Then the file closed.
Observed a healthy-looking bridge at one year, and a letter. Inferred that a recall would have changed it — maintenance does not abolish the risk.
Twelve roads out of that junction. One taken — and it is the only
one that removes you from the story.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+9:00. READ THE HEADLINE EXACTLY AS WRITTEN, IN THE FLAT VOICE OF A LETTER. THEN STOP FOR THREE SECONDS. This is the moment the format is built for: the room is waiting for a surgical rung and there is not going to be one. Do not soften it with 'of course we all do this'. Ask instead: who wrote the recall interval, in months, into the last implant discharge letter they signed? Then say the thing that is no longer written on the slide — the letter was written by someone doing their job, on a normal day, about a patient who was doing well, and that is what makes it the floor of this walk-back and not an error. WHY THAT ONE FILM MATTERS TWICE: the one-year radiograph is the single most predictive film anybody will ever take of this implant, and it is also the baseline every future diagnosis has to be measured against. WE HAVE JUST WALKED PAST A RUNG WITHOUT STOPPING — years two to seven, in which nothing happened, which is exactly the point. Say it in three sentences. One: the 2017 World Workshop definitions are the ticket barrier — health, mucositis, peri-implantitis, decided by bleeding, probing depth and radiographic bone loss compared with a baseline. Two: “in the absence of previous examination data” peri-implantitis is defined as bleeding or suppuration, probing depths ≥6 mm and bone levels ≥3 mm apical of the most coronal intraosseous part of the implant (Hakkers 2025, methods) — that is a definition written for patients whose records are missing, it is the one she gets, and it is not worse, it is just blind to everything in between. Three: what the seven years actually contain is a bridge fitted, a discharge letter and a radiograph eight years later — when it started, how fast it moved and whether it moved in one step or seven are now unobtainable. The natural history of her own disease is not recoverable, and that is a permanent loss, not an inconvenience. IF SOMEONE QUOTES YOU A PREVALENCE: the reported range across 22 primary studies is 6.68–62.3% at patient level and 4.5–58.1% at implant level (Afrashtehfar 2023) — a nine-fold spread at patient level and 13-fold at implant level, not because the disease varies that much but because the case definitions did. The first question is always which definition, not which population. And ask the room how many of them could produce the baseline radiograph for an implant they restored three years ago.
how far back
seven years earlier · the one-year review
12 roads out of this junction. 1 taken.
What was available at the one-year review
Taken — 1 of 12
Discharged to the referring practitioner
Available, not taken
Recall interval, in months, in the notesBaseline radiograph kept and labelledSix-point probing chart at the fitHome care and access for cleaningRisk profiling and supportive therapySmoking status in pack-yearsSmoking cessation referralGlycaemic control checkedHistory of periodontitis on the implant recordA named clinician who owns the reviewA prosthesis you can get a brush under
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+10:30. Thirty seconds, and it is the one junction in the hour where you point at the lit chip LAST. Read the eleven unlit roads first, slowly enough that the room hears what they cost: almost none of them costs theatre time, and most of them cost a line in a letter. Then drop your hand onto the single lit chip — discharged to the referring practitioner — and leave the pause. Nothing on this board was refused her. Eleven of them were simply never written down. Hold the count in your head, because the floor rung two slides on has the other half of it.
how far back
eight years earlier · before the first incision
Step back five · before she sat down ·
the floor of this walk-back
A history of periodontitis, and fifteen a day. Neither is on the implant record.
History of periodontitis and smoking head every risk list there is.
Neither of them reached the implant record, and neither set the recall interval.
Observed absent keratinised mucosa, OR 2.78 — still 2.08 under maintenance. Inferred a number for periodontitis or smoking. This hour has verified neither.
Every rung above this one is downstream of a question nobody asked. The
drill is eight years and four rungs away from where this went wrong.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+11:30. The floor. Two sentences and then move on to the junction, which is where the count lands. THE CHAIN, which used to be three boxes on this slide and is better said than read: risk was never assessed, so there was no reason to set anything other than a routine interval; a routine interval became a discharge, because nothing on the record argued against it; and then seven years of biofilm on a surface nobody could reach, with a disease that does not move down a slope. BE SCRUPULOUS ABOUT THE AMBER LINE. Foundation 7.3 lists history of periodontitis and smoking first among the risks and every guideline agrees, but this project has not verified an effect size for either of them. It has one for absent keratinised mucosa: OR 2.78 for peri-implantitis, 95% CI 2.07–3.74, GRADE moderate, and still OR 2.08 in patients on regular maintenance (Afrashtehfar 2023). And even that is contested — an umbrella review and a trial sequential analysis both put the effect of keratinised mucosa width on disease at low certainty (Ravidà 2022, Sabri 2025); it buys tissue health, not survival. So name the two risks and do not quote a number. “Two of the recognised risk factors were present and neither reached the implant record” is true. “Smoking triples it” is not something you can source from this hour, and the room will want you to say it. Saying so is worth more than the number would be.
how far back
eight years earlier · before the first incision
11 roads out of this junction. 0 taken.
What risk assessment could have been at the consent visit
Taken — 0 of 11
Available, not taken
Take a periodontal historyFull-mouth probing chart before you planSmoking, in pack-years, on the implant recordHbA1cKeratinised mucosa measured at the siteScrew-retained, so there is no cement to leaveContour the prosthesis for a brush, not a photographSet the recall interval from the risk, not the diaryWrite the risk into the consentSay the maintenance cost out loudDecide not to place
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+13:00. Thirty seconds, and it is the floor of the walk-back. Point at the TAKEN heading first, at the empty space under it, and say the number out loud: zero. Not one of these appears in her notes. Then put the two junctions side by side in words, because that contrast is the shape of the whole walk-back: one road of twelve at the one-year review, and none of eleven here. Every one of these is a question, a measurement or a sentence in a letter — there is not an instrument anywhere on the board. The last chip is the one to end on: decide not to place. It is a road, it was available, and it is the answer walk-back nine is about to spend fifteen minutes on.
how far back
eight years earlier · before the first incision
The ballot, answered · walk-back 8 of 9
You voted before you were told anything. The walk ended at C.
AThe implants were badly placed and the bridge could never be cleaned
BShe had peri-implantitis for years and it was caught far too late
CHer risk was never assessed, so there was nothing to set an interval fromwhere the walk ended
DShe was discharged at one year with no recall interval
Seven walk-backs have ended somewhere you could have stood with an instrument in your hand, so the room votes for the placement and the bridge. There is no surgical rung in this one — the file closed at one year because nothing on the record argued against closing it.
Observed bone gone at both, and a bridge that looked healthy at one year. Inferred that the placement or the bridge did it.
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+13:30. Fifteen seconds. Read the lit row out flatly and do not soften it. This is the one walk-back in the hour where most of the room is not slightly wrong but wrong about the kind of thing that went wrong, and they have to hear it said: nobody's hands did this. If anyone protests that the placement and the bridge really were poor, agree — and then ask what would have found that out, which is a recall interval nobody set, from a risk nobody wrote down. Do not relitigate the case. Go to the principle.
how far back
eight years earlier · before anything was decided
The floor · foundation 7 of 8 ·
biofilm and the peri-implant lesion
A biofilm disease is managed in the diary, not the theatre.
Larger, hotter, and it moves in steps rather than down a slope.
Every one of those is an argument about intervals — not about
instruments.
Observed a known driver, a predictor at one year, a 97% salvage figure. Inferred that a recall would have saved these two implants.
Eight of eight lit, and the board has been full since walk-back six. One left,
and it is the only one whose answer is don’t.
1Osseointegration1 walk-back
2Primary stability1 walk-back
3Bone and graft healing1 walk-back
4The soft tissue sealalso lights
5Load transfer without a PDL1 walk-back
6Three-dimensional position3 walk-backs
7Biofilm and the peri-implant lesionprimary here
8Anatomy and its limits3 walk-backs
← backwardsThe principleThe history not takenDischarged at one yearSeven years, no recallShe comes backThe outcome
+14:00. Let the headline sit, then advance once and let the board fill under it. Say out loud that nothing new lights — the 2 cards this walk-back touches were already on, and this is the first time in the hour that the scoreboard does not move. The room will read that as an anticlimax unless you name it: it is the first time foundation 7 is the WHOLE story rather than a passenger, and foundation 4 lights again because the seal she needed was the one thing the prosthesis design made impossible to clean. If you add one sentence, add this: seven of these nine walk-backs end at something you did with your hands, and this one ends at something you did with a keyboard, and it is the one that cost the most bone. Do not apologise for the slide being about admin. THE FULL PRINCIPLE, which is longer than the headline: peri-implantitis is biofilm-driven and it does not behave like periodontitis — the lesion is larger, it is hotter, and it moves in steps rather than down a slope; every one of those properties is an argument about how often somebody looks, against what baseline, and with what risk written down beside it. The operation is the thing you do when the interval has already failed, which is why the recall interval is a clinical decision with an evidence base and not an administrative one, even though it is taken by writing a letter. THE AMBER HALF, in full: observed — a disease with a known driver, a known predictor at one year, and a treatment with a 97% survival figure attached to it; inferred — that a recall interval would have saved these two implants. Nobody can show that and this walk-back does not claim it. What it claims is narrower and harder to argue with: the last point at which this outcome was cheap to change was a letter at one year, and the reason nobody thought to change it was that her risk had never been written down. This is the only one of the nine where the deepest rung is an administrative decision. That should bother you more than a broken drill. WHAT CHANGES ON FRIDAY — three things, none of them 'be careful', and none of them on the slide. One: write the recall interval into the discharge letter as a number of months, with the reason next to it; 'routine review' is not an interval. Two: take the one-year radiograph, label it BASELINE, and record six probing depths on the same day, because the 2017 definitions cannot be applied without them. Three: put smoking and any periodontal history on the IMPLANT record, not just the medical history, and let them set the interval. ABOUT THE JUNCTIONS, ONE LAST TIME: 41 roads across the three junction slides in this walk-back, 3 taken — and in the two that mattered, at the one-year review and at the consent visit, 1. The empty junction is not a graphic, it is the argument. Ask them which walk-back they would now put first if they were teaching this to a house officer. Then hand over to the coda: one more, her jaw broke, the walk-back goes back further than the day she sat down, and it is the only one where the right answer is to place nothing.
how far back
today · nine days after the fracture
Walk-back 9 of 9 · the coda ·
a composite patient — one published case
She is 74. Three weeks after a failed implant came out,
her mandible broke.
A · Sass 2021The day she presentedright side, through a fractured miniplate
Figure 1, and every photograph in this walk-back is the same
published patient, followed to six years — but she is not the patient being
narrated. Sass et al. 2021, case 1. Reproduced under CC BY 4.0.
Edentulous twenty-two years. Three lower
dentures, none of them wearable. Two years ago four implants went in to hold an
overdenture; one failed and came out at eight weeks. No trauma — she was
eating.
Observed a ribbon of bone with a fracture through it. Inferred nothing yet — not how much bone, not what was placed, not why.
This is the last of the nine, and it is the only one whose answer is
don’t.
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+0:00. Straight in — there is no seam slide on the coda, and the hour has earned the right to open on a photograph in silence. Put the radiograph up and say nothing for ten seconds. Most of the room will not have seen a mandible this thin. Give the history in the order written and then STOP: do not say the words 'atrophy' or 'pathological fracture' yet, because both of them are the answer. NOT ON THE SLIDE any more, and worth one sentence each if the room is slow to settle: the lower denture has been remade three times; the four implants went into the interforaminal region; the one that failed was removed at eight weeks and she fractured three weeks after that, on a sandwich. SAY ALOUD what the caption says in small print — this is ONE published patient, a 59-year-old whose extremely atrophied mandible fractured while she was eating, and she is not her. That declaration is not pedantry here; it is the pivot of the whole walk-back and you will detonate it on the next slide. EVERY PHOTOGRAPH FROM HERE IS THAT SAME WOMAN, in order — this radiograph, her face, her mouth before, and her mouth six years after a fibula. Say so once, now, and the room will stop wondering whose picture it is looking at and start following the story.
how far back
today · nothing revealed yet
Walk-back 9 · the ballot · anonymous, and compulsory
Before you are shown anything else: what went wrong?
ANothing should have been placed — the mandible was the disease—
BThe implants were longer than the bone could take—
CThe osteotomies removed what strength the mandible had left—
DOne implant had already failed, and the fracture ran through it—
Hold on to what you picked. Three of those four are about the
operation. The fourth is not, and it is where this walk-back ends.
Scan once — it follows the deck
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+1:30. Chase abstainers — ninth time of asking, and the habit is the mechanism. Expect D to lead: by walk-back nine the room has learned to look for the mechanical link, and D is the mechanically correct answer. Say nothing about the distribution. Poll id w9. If the phones cannot reach the laptop (NHS guest wifi very often isolates clients), press H and type counts with 1–9; 0 clears, V opens or closes voting. NOT ON THE SLIDE, and it does not need to be, because the closing slide is the hour's claim about itself: learners who attempt a solution before instruction outperform learners shown someone else's mistake, on transfer, immediately and at one week (Steenhof 2020, RCT). SAY ALOUD, once, the rule for this vote: you are not being asked who got this wrong, you are being asked at which point the outcome stopped being avoidable — and every previous walk-back had an answer that was a technique, where this one does not, which is what the vote is for. If you have been logging accuracy across the nine, this is the last data point.
how far back
two years back · the day the scan was reported
Step back one · two years ·
and the failed implant with it
That mandible never had an implant in it.
A · Sass 2021The same woman, the same day: the fracture has produced
marked facial asymmetry and an orocutaneous fistula — and there
is no implant anywhere in that jaw. Sass et al. 2021, case 1. Reproduced under CC BY 4.0.
Kablan says it plainly: “The fractures lines occurred at a
failed implant site.” True, and mechanically clean.
Observed less than 5 mm of bone from nerve to crest, in every case in the series. Inferred a rate from them — Sass is a selected series of three.
Hers broke with no implant, no osteotomy and no surgeon anywhere
near it. The implant gave your patient’s fracture an address. It is not
what made a fracture possible.
Kablan 2020 is paywalled and is not in Unpaywall — DOI
10.3290/j.qi.a43950. The sentence and the numbers are quotable; no figure of his is
reproduced here, and the quotation keeps his grammar.
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+3:30. Read the verbatim sentence out, including the grammar — it is quoted exactly and the slip is Kablan's, not ours. Then the headline, plainly, and STOP TALKING for a beat: this is the pivot and it only works as a surprise. The room has just voted for a mechanism involving an implant and neither of the two mandibles it has been looking at for five minutes had one. If anyone objects that these are selected cases, agree — they are, they are a case series of three, and that is exactly why the population figure below matters. THE NUMBERS, if challenged, and the habit is more important than either of them: 0.2% is implant-related spontaneous mandibular fracture in patients with implants in an edentulous mandible — the population figure. 1.5% is 2 of 98 patients, at 3 and 4 weeks, both managed conservatively — but that series is 132 inferior alveolar nerve transposition procedures over 10 years with 379 implants placed in one stage, a far more destructive operation in which the canal is opened, the nerve lifted out and the implants engage what is left. DO NOT PUT THOSE TWO ON ONE AXIS. Registrars quote 1.5% for atrophic mandibles constantly and it is the wrong number for that sentence; on a slide, in an exam or in a consent conversation, always say which operation the denominator came from. ALSO OFF THE SLIDE: both Sass fractures met the same criterion — a vertical defect of at least 1 cm over at least 5 cm, and less than 5 mm from the inferior alveolar nerve, or the base of the mandible, to the crest. One patient had spent twelve years failing to get a removable prosthesis that fitted, the other a decade and a half; the bone was going long before anyone proposed anything. AND IT HAS A NAME: Cawood & Howell grades what is left, in six classes, and the corpus uses it — Bär 2024 reports ridges as class V and class VI on the way to describing why an implant would not hold. The class definitions are the 1988 original and are NOT in this project's verified evidence, so do not put a cut-off on a slide; the discipline being taught is that you write a class and a millimetre figure in the notes, not that you can recite the boundaries.
how far back
two years back · the moment the drill went in
Step back two · the moment the drill went in
The answer to bone you cannot grip is not a longer implant.
Six millimetres is two cortices and nothing betweenno cancellous bone for thread purchase
So the only grip available is bicorticalengaging a lower border that is already the lower border
And every osteotomy is a notch through most of the depthpermanent, and empty if it fails
Observed a crest the floor of the mouth stands above. Inferred that the osteotomy is the stress riser — nobody here has measured it.
A · Sass 2021The same woman, before her reconstruction:
the floor of the mouth stands above the crest. There is no ridge here to
put anything into. Sass et al. 2021, case 1. Reproduced under CC BY 4.0.
Remote anchorage is an answer, not an exemption. A subperiosteal framework
on an atrophic maxilla for twenty years came back as a T4a squamous cell
carcinoma.
Watanabe 2022 is named and quoted here and no figure of it is
reproduced — every photograph in this walk-back is one patient. DOI
10.1186/s40729-022-00409-3.
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+7:00. Foundation 2 lights here and the room should feel it arrive from the opposite direction to walk-back 1, where an implant fell into a sinus for the same reason. Let the chain run without commentary; the third link is the one to point at, because the notch is permanent whether the implant integrates or not. THE TREE IS OFF THE SLIDE and it is worth a sentence: §2.5 names three places you cannot get primary stability — the fresh extraction socket, grafted bone and the Type IV posterior maxilla — and gives all three the same answer, remote anchorage: go and get your grip somewhere else. The severely atrophic mandible is the fourth case and the answer does not change. That extension is mine; the three named sites and the answer are the tree's. WATANABE IS NOT A SCARE STORY AND MUST NOT BE DELIVERED AS ONE. Twenty seconds, in this order: it was placed on a severely atrophied maxilla in 1999 because there was not enough bone for anything else; one retaining screw fell off almost immediately; he was followed for five years and then not seen again; he came back in 2019 with a T4a squamous cell carcinoma around it, with no smoking and no alcohol history. The authors' own stated hypothesis, offered as one, is chronic mechanical irritation plus untreated peri-implantitis. It is here because it is the honest caveat at the exact moment you introduce remote anchorage, and because it primes the handover on the last slide — not because subperiosteal implants are wicked.
how far back
before she sat down · the history
Step back three · before she sat down ·
past the consent form
Twelve things could have changed it.
One of them was written down.
None of these arrive through the drill. They arrive through the
history, and the history is taken before she sits down — or it is not
taken.
Observed excellent case reports of osteonecrosis around implants. Inferred a rate — this project has no verified denominator, so the gap is on the slide.
Favia 2015 carries a no-derivatives licence: the numbers are
quotable, and no figure of it is reproduced or cropped here.
Thirty-three months of monthly intravenous zoledronate. Then necrotic bone
around four implants placed before the drug and sound when it
started — and, invisible on every image,
metastatic breast cancer in about 5% of the sample. That is why it goes
to histology.
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+10:00. The taxonomy is the junction slide after this one, so hold the counts until then. The Favia case is the one to say slowly: a woman of 66, breast cancer diagnosed in 2005, then bone metastases and monthly intravenous zoledronate for thirty-three months. A lesion that looked exactly like peri-implant osteonecrosis — pain, pus discharge and paraesthesia of the right inferior alveolar nerve — around four implants in the RIGHT MANDIBLE, at 3.1, 4.1, 4.4 and 4.6. Every one of those implants had been placed in 2008, more than six months BEFORE the drug was started, and all were radiologically well osseointegrated when it began; that is precisely what makes it a non-implant-surgery-triggered lesion, and it is the sentence to lean on — nobody's drill did this, and three of her other implants were fine throughout. Then the part that matters: about 5% of the resected specimen was metastatic breast cancer, found by histology alone, with nothing on the clinical picture or the imaging to suggest it and only four such cases in the literature at the time. It is the reason 'send it for histology' is not defensive medicine. Resist the urge to add a rate to the MRONJ beat — we do not have one and saying so is the teaching point. THE NUMBER THAT CAME OFF THE SLIDE: implant failure in proton-pump-inhibitor users against non-users, 12.0% vs 4.5%, 3559 implants in 999 patients — a drug class nobody asks about, reported inside a review of something else (Mauprivez, in Alla 2023). THE RUNG WE WALKED PAST, and it is worth thirty seconds because it is the decision itself: there were eight roads out of that junction and she was offered one — a complete denture. Six of the eight are not available to an edentulous mandible at all, which is the point: by the time she sat down the junction had two roads. In the two published patients the denture road had been run to exhaustion — twelve and fifteen years trying to have a prosthesis made that fitted, and neither succeeded. That is a real argument for doing something. It is not, by itself, an argument for doing this. And we cannot claim our composite patient was never offered an alternative — the notes are invented. What is claimable is the general case: the non-implant option is almost never written down as an offer, and if it is not written down it did not happen. Ask by show of hands who has ever written 'declined a further denture' in a set of implant notes, and do not comment on the count. The full eight roads are in the print appendix.
how far back
before she sat down · the history
12 roads out of this junction. 1 taken.
The patient-level limits — and what the notes actually recorded
Taken — 1 of 12
Consent and documentation
Available, not taken
SmokingGlycaemic controlHistory of periodontitisAntiresorptives and MRONJIrradiated boneParafunctionChronic kidney diseaseExpectation and psychological assessment
On the map, not in the evidence
Hyperbaric oxygenImmunosuppressionAnticoagulation
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+11:00. Thirty seconds, and the whole of it is the arithmetic — SAY BOTH NUMBERS OUT LOUD, in this order: twelve things could have changed this, and one of them was written down. Do not let the slide say it for you; the contrast in the counts is the teaching and it only lands spoken. Point at the lit chip as you say the second number — consent and documentation, the only entry on the whole list with a box on a form, which is exactly why it is the only one that got done. The dashed chips are the station list's own doubts and not mine: hyperbaric oxygen, immunosuppression, anticoagulation. One chip lit out of 12 is uncomfortable and it should be. If you have time for one more pair of numbers, give them the junction we walked past: eight roads out of the decision to place at all, and she was offered one.
how far back
before she sat down · the history
The ballot, answered · walk-back 9 of 9
You voted before you were told anything. The walk ended at A.
ANothing should have been placed — the mandible was the diseasewhere the walk ended
BThe implants were longer than the bone could take
CThe osteotomies removed what strength the mandible had left
DOne implant had already failed, and the fracture ran through it
The fracture did run through the failed implant site — Kablan says so, and it is mechanically right. Both published mandibles broke with no implant, no osteotomy and no surgeon anywhere near them: the implant gave the fracture an address, not a cause.
Observed a fracture through a failed implant site, and two mandibles that broke with no implant at all. Inferred that the implant caused it.
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+12:00. Fifteen seconds, and the ninth and last time you do this. Read the lit row out and let the room see how far the answer sits from the three that are about the operation. Say the credit where it is due: a room that voted for the failed implant site has been reasoning mechanically for an hour and has got the mechanism right — it is the depth that is wrong, and the pivot two slides back is why. Do not gloat, and do not explain it twice; the principle says it properly on the next press.
how far back
before she sat down · before any of it
The floor · foundation 8 of
8 · anatomy and its limits
The atrophy was the disease, not the consequence.
Losing the teeth was the first thing it did, not the reason it
happened. The mandible had been shrinking for two decades and was still shrinking
underneath the implants.
Observed less than 5 mm, measurable years before either mandible broke. Inferred that asking earlier would have changed this one — no series can show that.
“Can I place implants here” was the wrong question. The
right one was: what is still happening to this mandible, and does anything I do
make it happen faster?
← backwardsThe principleBefore she sat downThe decision to placeThe bone that was leftThe implant that failedThe fracture
+13:00. Let the headline sit. This is the sentence the coda exists to produce and it should feel like the end of an argument rather than the end of a case — so do not summarise the case again underneath it. If you say one extra sentence, say that the assessment IS the intervention here, and that it is the only walk-back in the hour of which that is true. WHAT 'DON'T' DOES AND DOES NOT MEAN, and say it if anyone bristles: it is not an argument that nothing may ever go into an atrophic mandible, it is an argument about the order of the questions. WHAT CHANGES ON FRIDAY — say all three, they are not on the slide. One: measure it and write the class down; 'very atrophic' is not a measurement, and a height in millimetres at each planned site plus a Cawood & Howell class belongs in the notes before the word implant is used. Two: write down what you would have done instead and that it was offered — rung 0 belongs in the record as an option, not as an omission. Three: if an implant in an atrophic mandible fails, treat that site as a fracture risk for the next month rather than as a re-implantation site. Then hand over to the counterweight, or the room leaves thinking the answer is that nothing can be done.
how far back
The counterweight · what it costs when
don’t arrives too late
The answer to a broken atrophic mandible was a
free flap.
A · Sass 2021Six years onthe whole corpus rebuilt, two implantsA · Sass 2021The same mouthkeratinised gingiva, no bleeding on probing
The same woman, six years on — figures 1 and 2. Sass et al. 2021, case 1. Reproduced under CC BY 4.0.
Free vascularised fibula, implants at six
months, a locator-retained overdenture. Nothing smaller worked.
So don’t is not nihilism; the ceiling is high. The price of
reaching it late is microvascular reconstruction in a 59-year-old,
with a donor site that can cost sensory loss or ankle instability.
Observed probing depths of 2 mm and no bleeding, six years after a fibula. Inferred a rate — three cases, and the authors call the indication rare.
+15:30. The humane beat, and it stops the coda ending on a wagging finger. Two things the room should take: the outcome after reconstruction is genuinely good, and the operation that produces it is not one most of them will be doing. If you are running short, this is the slide to compress — but do not cut it, because without it the answer 'don't' sounds like 'nothing can be done'. NOT ON THE SLIDE: the second case was a fibula with a skin paddle, six implants at ten months, a screw-retained full-arch bridge and no resorption at five years; both patients spent ten days in hospital. The distraction figure is worth quoting in the authors' own words — at least 5 mm of intact height is needed 'to ensure that the bone is not resorbed or fractured', which is to say the operation you would reach for needs more bone than she has. And the honest denominator: this is a case series of three, from a department that did 460 microsurgical procedures in the same period, and the authors call the indication rare. Donor-site morbidity in general can cost sensory loss, ankle instability or contracture of the great toe; none was reported permanent in these two.
how far back
Walk-back 9 of 9 ·
eight of the eight foundations lit
Nine complications.
Eight foundations.
Not one of them ended at the drill.
1Osseointegration1 walk-back
2Primary stabilityalso lights
3Bone and graft healing1 walk-back
4The soft tissue seal4 walk-backs
5Load transfer without a PDL1 walk-back
6Three-dimensional position3 walk-backs
7Biofilm and the peri-implant lesion2 walk-backs
8Anatomy and its limitsprimary here
The further back you have to walk, the less it is about the drill and the more
it is about the patient.
The bar along the top is the whole hour. It
started at thirty seconds and an implant in a sinus, and it ends before this patient
sat down. Nobody was told that sentence.
Say this out loud. Everything past this point — zygoma, pterygoid,
fibula — is the same eight foundations at higher ambition, and that is
the climb, after the break. The ladder has nine rungs; this
hour reached rung 0. The other eight are at
11:20.
Observed nine complications, and not one caused by not knowing a taxonomy. Inferred that the ladder does not matter — it is the 11:20 session, not an omission.
+18:00. THE NUMBERS ARE OFF THE SLIDE so the two photographs can be seen: six years, 2 mm of bone loss at one implant, probing depths of 2 mm, no bleeding. Ramus and symphysis yield no more than 5 mm of vertical gain; iliac crest resorbs 30-47%; distraction needs 5 mm of intact height to begin. THE HANDOVER IS MANDATORY AND IT IS SPOKEN — 15-coverage-audit.md §3. Read the crimson box aloud, in those words, before anyone stands up: everything past this point — zygoma, pterygoid, fibula — is the same eight foundations at higher ambition, and that is the climb, after the break. Without it the room's honest impression at the end of this hour is that implantology stops at the alveolus. The nine things it covers, if you want to name more than three: zygomatic — intrasinus · zygomatic — extramaxillary / ZAGA · quad zygoma · pterygoid · tuberosity · subperiosteal — CAD/CAM · transplanted bone — fibula, DCIA, scapula · jaw in a day · craniofacial — auricular, orbital, nasal. WHY THEY ARE NOT IN THIS HOUR, by design, and worth one sentence: a walk-back is a causal instrument, excellent at this decision caused that outcome and structurally incapable of arriving at a ladder, because no complication is ever caused by not knowing one. THEN THE ONE SENTENCE: say it once, slowly, pointing at the depth meter, and do NOT explain it. It is the payoff of the ordering, not a claim you argue for; the hour started at thirty seconds and ends before this patient sat down, and nobody in the room was told the sentence in advance. If you have been logging the commit votes across the nine, show the accuracy curve here — it is the day's own claim, made falsifiable, and it takes fifteen seconds.
how far back
The legend · the scale bar for all nine walk-backs
· read this before the first photograph
Survival asks one question: is it still in the mouth?
Success asks whether a named set of conditions has been met — bone
level, no mobility, no pain, no suppuration, a prosthesis that works.
Quoting a survival figure as if it were a success figure is the commonest
sleight of hand in the implant literature.
97.2%survived — 8 of 284 implants
failed
61.4%met success — “full
success, optimal health”
One cohort. Same implants, same ten years,
same paper, scored against the ICOI Pisa criteria. A quarter sat in
compromised survival: alive, and not well. 27.5% of implants were lost
to follow-up, so the success denominator is the smaller one.Degidi 2015 · doi:10.1111/clr.12642 · PMID 26096295
When someone quotes you a percentage — a rep, a paper,
a colleague, one of these nine walk-backs — there are two questions before
you believe it. Which definition? And whose criteria?
A CARD, not a slide — forty seconds, and it is the only card before a photograph. Read the two numbers and say nothing else about them: the room supplies the outrage on its own. Do NOT start on peri-implantitis definitions here — foundation 7 gets its own card at walk-back 6. If you want one line of your own, make it the last one: which definition, whose criteria.
how far back
The legend · part two · every number on this card
carries its DOI or PMID
“Success” is whatever
the paper said it was.
Five sets in common use. They do not measure the same
thing. Read across a row — and look at the gaps.
Albrektsson & Zarb 1986judges the implant
Smith & Zarb 1989adds the prosthesis
Buser 1990pass or fail, fixed in advance
ICOI Pisa 2007 · Misch 2008a health scale, not a verdict
What papers actually useacross studies at five years or more
Immobile
when tested clinically
tested individually
yes
mobility is failure
yes
No radiolucency
none on a radiograph
none distinct
none continuous
—
yes
No pain or altered sensation
incl. neuropathy, paraesthesia, mandibular canal
pain or discomfort
incl. foreign-body sensation, dysaesthesia
none on function
yes
No infection
yes
yes
none recurrent, with suppuration
no exudate
—
Bone loss
< 0.2 mm/yr after year one
≤ 0.2 mm/yr after placement
left out on purpose
graded — 2 mm · 2–4 mm · > 4 mm
> 1.5 mm fails
The prosthesis
—
the crown looks right — to the patient and to the dentist
restoration possible
—
—
A minimum rate
85% at 5 yr · 80% at 10
85% at 5 yr · 80% at 10
—
no threshold — four grades
—
Int J Oral Maxillofac Implants 1(1):11–25 · PMID 3527955 · Table 1 is a bitmap, read by eye
J Prosthet Dent 62(5):567–572 · our copy is a back-translation from a Japanese presentation — paraphrased, not quoted
The gaps are decisions, not oversights.
Buser left bone loss out because the ITI implants he was reporting barely
lost any. And only Smith & Zarb ask whether the crown looks right
— in 1989, this day’s argument already in the literature, and
then dropped. Lee et al. 2024 ·
doi:10.32542/implantology.2024013
Forty seconds, and do not read the table. It is a reference object, not a script — the room reads a table far faster than you can say it, so give them four seconds of silence to look and then point at the THREE CELLS that carry the argument. (1) The prosthesis row. Four of the five sets are blank. Only Smith & Zarb ask whether the crown looks right, and they ask it of the PATIENT as well as the dentist — in 1989. That is this day's whole argument, already in the literature, and then largely dropped. (2) Buser's bone-loss cell. Empty on purpose, because the ITI implants he was reporting barely lost any. A criterion missing from a set is a decision. (3) The bottom row. Two sets demand 85% at five years; two do not set a threshold at all. Same word, different bar. SAY THIS, it is off the slide: the zygomatic literature admits it outright — ‘there are no clinically applicable criteria for success ... most studies consider success to be the survival of the implants placed’ (Van den Borre 2023, doi:10.3390/jpm13020297). TWO PROVENANCE POINTS, if anyone presses on the numbers. Albrektsson's 0.2 mm sits in a table that is a BITMAP in the source PDF — transcribed by eye, not machine-matched against the file's own characters, so it is high confidence and not proof, and the slide says so. Smith & Zarb reaches us as a back-translation from a Japanese presentation of the JPD paper, so it is paraphrased rather than quoted. Both are on the slide; you do not need to raise them, but do not let either pass as verbatim if challenged. If a registrar asks which set to use in their own logbook: ICOI Pisa, because it is the only one that lets an implant be alive and not well. Then go to the first photograph and do not mention criteria again until 10:00.
how far back
The close · every case, and the sentence it ended on ·
photograph this one
Nine complications. Nine conclusions.
1It's gone30 seconds · Implant displaced into the maxillary sinusMechanical grip buys the time biology needs.
2Her tongue is against her palate10 minutes · Life-threatening sublingual haematomaThe measurement was vertical. The danger was horizontal.
3It came out in my hand6 weeks · Failure to integrateBone tolerates it. It never bonds to it.
4The membrane is showing7 months · The graft that failedRegeneration is a space, not a substance.
5You can see the metal18 months · Recession in the aesthetic zoneThe seal is the only barrier. It is only as thick as the bone under it.
6It bleeds when she brushes3 years · Peri-implantitis at an anterior maxillary implantEverything downstream is decided before the drill goes in.
7The screw snapped4 years · The mechanical failureThere is no ligament. The load ends in bone, or it ends in metal.
8Nobody ever saw her again8 years · The one that is not the surgeon’s faultA biofilm disease is managed in the diary, not the theatre.
9Her jaw brokebefore she sat down · The coda — the only walk-back whose answer is don’tThe atrophy was the disease, not the consequence.
THE SLIDE TO PHOTOGRAPH — say so, and give them ten seconds before you speak. Do NOT read the column. Every sentence in it is one they have already heard land at the end of a case, so reading it back is the one thing that makes it feel like a summary rather than a return. Point at two: the one this particular room needs, and the last one. Then say the sentence the ordering produced — the further back you have to walk, the less it is about the drill and the more it is about the patient — and stop.
how far back
Everything this hour cited · 45 papers ·
photograph this, or take the handout
Do not read it. Say it exists, say the handout carries it with the DOIs live, and move. Anyone who wants a paper can photograph the slide; anyone who wants all of them has the manual. Every one of these is either open access or reachable through OpenAthens from the DOI.
how far back
Everything this hour cited · 45 papers ·
photograph this, or take the handout
Do not read it. Say it exists, say the handout carries it with the DOIs live, and move. Anyone who wants a paper can photograph the slide; anyone who wants all of them has the manual. Every one of these is either open access or reachable through OpenAthens from the DOI.
how far back
The floor of the hour · and the only claim it makes about itself
You voted nine times before you were told anything.
One30 sec—
Two10 min—
Three6 wks—
Four7 mths—
Five18 mths—
Six3 yrs—
Seven4 yrs—
Eight8 yrs—
Ninebefore—
Share of the room that picked the rung the walk-back
reached. Live, from your own votes. If the bars are empty the server is
not running.
Being wrong early was the mechanism, not an embarrassment.
The evidence is one randomised trial, and you just spent an hour inside it.
Generate
Learners who attempt a solution
before instruction outperform learners shown someone
else’s mistake
Steenhof 2020, RCT
Which is the whole reason this hour ran backwards.
Observed how a room of twenty voted, nine times, on nine composite patients. Inferred that any of it changes what you do at 14:00.
AK1.3
Steenhof et al. 2020, RCT
Advance to this only if the poll server has been running all hour. Read the shape, not the numbers — the claim is that the line climbs, and if it does not, say so out loud. That is the point of putting it on the screen. The Steenhof finding holds on transfer, immediately AND at one week — say that, it is what makes watching the losing arm. And say the caveat out loud too: one room, one morning, no control arm; the same rule we applied to every case report applies to the lecture that used them. Then hand over: the ladder handout is the take-away, the stations start at 14:00, and rungs 1 and 3 are the two you can stand on with a jaw model.
OMFS Implants Study Day · 7 August 2026 · 09:00 Foundations — Foundations — nine walk-backs
The roads not taken
OMFS Implants Study Day · 7 August 2026 · 09:00 Foundations — every option that existed at each decision rung of the nine walk-backs. On the slide you saw a count and the roads actually taken; this is the rest. ● taken · ○ existed, not taken · ◇ not evidenced.
Walk-back 1 · 30 seconds · it's gone
grip — 4 roads, 1 taken. ○ Insertion torque, in Ncm · ○ Resonance frequency analysis — ISQ · ○ Periotest · ● Tactile judgement at the driver
Walk-back 2 · 10 minutes · her tongue is against her palate
airway — 14 roads, 2 taken. ● Stop, and get help into the room · ● Secure the airway · ○ Oral or nasal intubation · ○ Percutaneous tracheostomy, no prior intubation · ○ Laryngeal mask or Guedel airway · ○ Compression at the perforation · ○ Observation in a monitored bed · ○ Transfer — theatre and ITU standing by · ○ Incision and drainage of the haematoma · ○ Extraoral exploration and vessel ligation · ○ External carotid ligation · ○ Angiographic embolisation · ○ Leave the implant, remove it, or abandon · ◇ Reverse the anticoagulant
timing — 15 roads, 3 taken. ● Immediate — Type 1, the day of extraction · ○ Early — Type 2, four to eight weeks · ○ Early — Type 3, twelve to sixteen weeks · ○ Late — Type 4, healed ridge · ○ Spontaneous healing, then delayed placement · ○ Socket / alveolar ridge preservation · ○ Atraumatic extraction — periotome, piezo · ○ The buccal gap grafted · ● The buccal gap left to fill · ○ Simultaneous GBR at placement · ○ Staged GBR, then place · ○ Flapless · ○ Full-thickness flap · ○ Submerged healing, second-stage uncovering · ● Immediate provisional, non-occlusal
tissue — 14 roads, 3 taken. ● Nothing — at the time of placement · ● Connective tissue graft — envelope · ● Roll flap and pedicle techniques · ○ Free gingival graft · ○ Coronally advanced flap plus graft · ◇ Xenogeneic soft tissue matrix · ○ Vestibuloplasty · ○ Phenotype modification at placement · ○ Submerged healing · ○ Open transmucosal healing · ○ Provisional to shape the emergence · ○ Recontour the crown, open the angle · ○ Pink ceramic camouflage · ○ Explant and start again
Walk-back 6 · 3 years · it bleeds when she brushes
design — 16 roads, 1 taken. ○ Single crown · ○ Implant-supported FPD · ○ Cantilever FPD · ○ Full-arch fixed — metal-acrylic · ○ Full-arch fixed — monolithic zirconia · ● Overdenture on two implants · ○ Overdenture on four implants or a bar · ○ Conventional complete denture · ○ Cantilever length limits · ○ Cross-arch stabilisation · ○ Occlusal scheme — mutually protected or group function · ○ Cusp inclination and table width · ◇ Implant-protected occlusion · ○ Parafunction management and nightguard · ○ Occlusal adjustment at review · ◇ Progressive loading
loose — 13 roads, 4 taken. ○ Retighten it to torque · ○ Replace the screw · ● Replace the locator insert · ● Adjust the attachment · ● Reline the denture · ● Trim the overgrown tissue · ○ Radiograph it and record the pattern · ○ Check and adjust the occlusion · ○ Verify passive fit · ○ Ask about parafunction · ○ Add an implant or change the design · ○ Nightguard · ○ Send it back to whoever planned it
Walk-back 8 · 8 years · nobody ever saw her again
now — 18 roads, 2 taken. ● Non-surgical debridement · ○ Adjunctive antimicrobials · ○ Surface decontamination — mechanical · ○ Surface decontamination — chemical · ○ Surface decontamination — laser · ○ Access flap surgery · ○ Resective surgery · ○ Implantoplasty · ○ Regenerative surgery · ○ Combined resective and regenerative · ○ Supportive maintenance after treatment · ○ Refractory disease · ◇ Explantation — reverse torque · ◇ Explantation — trephine · ◇ Explantation — piezo · ○ The ridge afterwards · ◇ Re-implantation · ● What you tell the patient
risk — 11 roads, 0 taken. ○ Take a periodontal history · ○ Full-mouth probing chart before you plan · ○ Smoking, in pack-years, on the implant record · ○ HbA1c · ○ Keratinised mucosa measured at the site · ○ Screw-retained, so there is no cement to leave · ○ Contour the prosthesis for a brush, not a photograph · ○ Set the recall interval from the risk, not the diary · ○ Write the risk into the consent · ○ Say the maintenance cost out loud · ○ Decide not to place
year — 12 roads, 1 taken. ○ Recall interval, in months, in the notes · ○ Baseline radiograph kept and labelled · ○ Six-point probing chart at the fit · ○ Home care and access for cleaning · ○ Risk profiling and supportive therapy · ○ Smoking status in pack-years · ○ Smoking cessation referral · ○ Glycaemic control checked · ○ History of periodontitis on the implant record · ○ A named clinician who owns the review · ○ A prosthesis you can get a brush under · ● Discharged to the referring practitioner
Walk-back 9 · before she sat down · her jaw broke
risks — 12 roads, 1 taken. ○ Smoking · ○ Glycaemic control · ○ History of periodontitis · ○ Antiresorptives and MRONJ · ○ Irradiated bone · ◇ Hyperbaric oxygen · ○ Parafunction · ◇ Immunosuppression · ◇ Anticoagulation · ○ Chronic kidney disease · ○ Expectation and psychological assessment · ● Consent and documentation
