The Kinematic Thesis

Front matterPreface: who wrote this, and why that matters

I am not a surgeon. I hold an MD and I left an orthopaedic residency in my fourth year, which means I have deep clinical training and no clinical authority whatsoever. I have spent the twenty years since at the boundary between operating rooms and the companies that build what goes into them — most consequentially building the orthopaedic vertical at a surgical navigation company from nothing to roughly two hundred and fifty million dollars, and as a named inventor on a handful of image-guided navigation patents that are now expired or abandoned and therefore public property.

That standing is a genuine limitation and I want it stated on the first page rather than discovered on the fortieth. Nothing in this book should be read as clinical guidance. I have not operated on anyone in two decades and I will not pretend that reading the literature is the same as owning the outcome.

But the standing is also the reason the book exists, and I think that is worth being direct about. A knee surgeon reads knee literature. A spine surgeon reads spine literature. Both read deeply, and both read almost exclusively within their own field, because there is more than enough there to fill a career. Someone who sells navigation into all of those rooms reads across them out of commercial necessity, and hears the same conversation happening in four places at once with none of the participants aware of the others.

That is the observation this book is built on. It is not a clinical insight. It is a structural one, and it is available precisely to someone standing where I stand.

What this is

An argument, in four parts, that a single methodological error has been discovered and repaired independently in at least three orthopaedic subspecialties, that the repair is the same repair each time, and that naming it lets each field borrow what the others have already learned instead of arriving at it separately over another decade.

Eleven joints are worked through as evidence — seven built to state the argument, and four built afterwards to test it. One broke part of it, and two more forced corrections the original argument had no way to anticipate. Each has a companion instrument on this site that computes what the chapter describes, so nothing here is an assertion you have to take on faith — you can put numbers in and watch the argument behave. Every one of those instruments is a design instrument. None is a cleared device, none has been validated, and several rest on assumptions I have flagged in place rather than smoothed over.

What this is not

It is not a guideline, a systematic review, or a substitute for the primary literature. Where I cite a figure I have tried to name the source and its strength in the same breath, because a number with a citation and a number with a citation and a caveat are different objects and the difference matters. Chapter fifteen separates what is anchored from what I assumed to make the models run, and chapter sixteen states what would prove the whole thesis wrong.

Part I · The shared error

Chapter oneThree fields, one mistake

Between roughly 1970 and 2010, three orthopaedic subspecialties each settled on a universal anatomical target. Each target was reasonable, each was derived from population data, and each was applied to individual patients as though the population and the individual were the same thing. Each has since been shown to fail, in the same way, for the same reason.

The knee took the neutral mechanical axis. Cut both bones perpendicular to the mechanical axis, land the limb at zero, and the load distributes evenly. It was clean, it was teachable, and it produced an implant survivorship curve that justified itself for forty years. It also produced a persistent minority of patients who were mechanically perfect and functionally unhappy, and a body of work eventually established that most people do not have neutral knees to begin with. Correcting a constitutionally varus knee to neutral is not restoring it. It is imposing an average on someone who never had it.

The hip took the Lewinnek safe zone. Forty degrees of inclination, fifteen of anteversion, ten either side. Also clean, also teachable, and also derived from a real observation. It fails in a way that ought to have ended the argument on its own: when Abdel and colleagues looked at two hundred and six dislocated hips, fifty-eight per cent of them had the cup inside the safe zone. Other series report fifty-seven and fifty-eight per cent. The zone does not separate the hips that dislocate from the hips that do not, because a cup is aimed once in one position and the patient then stands up and sits down for the rest of their life.

The spine took neutral sagittal alignment. Restore lumbar lordosis to match pelvic incidence, bring the sagittal axis over the pelvis, and the patient stands efficiently. Again reasonable, again population-derived. And again it fails, but here it fails in the cruellest way available: the failure is caused by doing it well. Proximal junctional kyphosis rises from about eighteen per cent in young adults to about fifty in the elderly, and when Lafage and colleagues examined who it happened to, the patients who developed it had been overcorrected relative to age-adjusted goals. The neutral target is not merely imprecise in an older spine. Achieving it is a mechanism of harm.

Three fields, three targets, three failures. The failure is not that the numbers were wrong. It is that they were universal.

What each field did next

Here is the part that interests me. All three fields fixed it, all three fixed it the same way, and all three did so without reference to each other.

Read the last column downward. Constitutional alignment. Spinopelvic behaviour. The patient's decade. Their premorbid glenoid. Their hindfoot. Five subspecialties, five different words, one idea: the reference is the person on the table, not the population they were drawn from.

I do not think any of these fields borrowed from the others. The knee literature on constitutional alignment does not cite the hip literature on spinopelvic mobility. The spine work on age-adjusted targets does not cite the shoulder work on premorbid glenoid reconstruction. They converged because they were all making the same mistake and the mistake has one exit.

Why naming it is worth something

If this is one error rather than five, then each field is sitting on solutions the others have not yet reached for, and the transfer is cheap. Three examples of what becomes available the moment you look sideways:

• The hip has a mature two-position protocol — standing and seated films — for measuring whether an adjacent segment will move. The ankle has the same problem with the subtalar joint and solves it with a wooden block. Neither field describes what it is doing in terms the other would recognise, and both are answering the identical question.
• The spine has done the most rigorous work anywhere in orthopaedics on why a target must be indexed to the patient's age. No other joint in this book adjusts anything for age. It is not obvious that they should not.
• The hip has learned to compute a target and then discover it can safely decline the bailout — dual mobility use in its highest-risk group fell from every case to roughly a third once targets were properly planned. That is a general result about what good planning buys, and it is stated nowhere outside the hip.

The rest of Part One states the method these fields converged on, in language meant to travel between them. Part Two works it through seven joints. Part Three names three structural features that only become visible once you have seen the pattern several times. Part Four is where I try to take the argument apart.

Chapter twoWhat an envelope is

An envelope is what a joint does, measured across the range in which it does it. It is not a shape and it is not a single number. Every instrument in this book is built on one, and the first discipline the method imposes is refusing to accept a static measurement as a description of a moving thing.

A static measurement asks: what does this look like right now, in this one position, usually supine and usually unloaded. An envelope asks: what does this do across the positions the patient actually occupies, and under the loads they actually apply. The distinction sounds pedantic until you notice how much of orthopaedic planning rests on the first kind of measurement and how much of orthopaedic failure is explained by the second.

The knee's envelope is the compartment opening under a standardised stress, sampled across flexion rather than measured once in extension. The hip's is the functional tilt of the pelvis across posture, which requires the patient to be standing in one radiograph and seated in another. The ankle's is what the hindfoot gives back when you remove the forefoot driver. The shoulder's is not a motion at all but a distance — how far the glenoid has eroded from where it started. The spine's is sagittal alignment read against the patient's own decade.

They are not all the same kind of measurement, and I want to be honest that "envelope" is doing some work as a metaphor in a couple of those cases. The shoulder's is a displacement, not a range. The hand's, as chapter ten concedes, is not an envelope at all. But the operative question is identical in every case: what does this joint do that a single static view of it would not have told me?

The second look

In practice an envelope requires a second observation that most workflows do not currently take. This is the concrete cost of the method and it is worth naming plainly, because it is where adoption actually stalls.

That last row is deliberate. The spine's second look costs nothing, requires no equipment, and is already in the chart. It is also, on the evidence in chapter eleven, the single highest-yield second look in this book. Not every improvement in this method is expensive.

Chapter threeWhose body is the reference

Every measurement is a comparison against something. The question that organises this entire book is what that something should be, and the answer the five fields converged on is: something belonging to this patient, which existed before the disease and did not change when it arrived.

Call it the fixed reference. It has three properties. It belongs to the individual rather than the population. It is unaltered by the pathology you are treating. And it is unaltered by the operation you are about to perform. Anything with those three properties is a candidate; anything missing one of them is not.

The reference that is not anatomy

The spine's answer deserves a moment because it is the least expected and, I think, the most important. Nothing about a seventy-eight-year-old's radiograph tells you their target. Their pelvic incidence is fixed and knowable, and it still does not tell you where to put their lordosis, because the correct relationship between the two changes across a lifetime. The reference that resolves it is a number in the demographics field.

Once you have seen that, the question generalises uncomfortably. If the right sagittal target depends on the patient's decade, does the right knee laxity? Does the right cup position? A seventy-five-year-old and a forty-year-old with identical spinopelvic films almost certainly do not want identical hips, and I am not aware of a body of work that indexes cup planning to age the way spine planning now does. I raise it as an open question in chapter sixteen rather than as a claim.

The failure mode of a borrowed reference

When the reference is borrowed from a population rather than taken from the patient, the error is not random. It is systematic and it is largest exactly where the patient is furthest from the mean — which is to say, in the patients whose deformity brought them to you. The population target is most wrong for the people who need surgery most. That is not a small statistical inconvenience; it is an inversion, and it explains why each of these universal targets performed acceptably in aggregate for decades while failing a consistent and identifiable minority.

Chapter fourPosition first, escalate on the residual

Once you have an envelope and a reference, a sequence follows, and the sequence is the practical core of the method. Move what is cheap and reversible before you change what is expensive and permanent. Then measure what is left, and let that — only that — decide the escalation.

In every joint in this book the same three steps appear:

1. Measure the envelope against the fixed reference. This gives you a deviation, not a diagnosis.
2. Absorb what positioning can absorb. Component angles inside published boundaries. Cup anteversion inside a functional window. Corrective reaming inside a safe limit. Intra-articular balancing. These are the moves that cost little and can be reconsidered.
3. Escalate only on the residual. Whatever positioning could not absorb is the argument — the only argument — for changing the implant, adding an augment, reaching below the joint, or accepting a constrained construct.

Stated that way it sounds obvious. It is not what is usually done. The usual sequence chooses the construct early, often before the patient is on the table and sometimes at the point of scheduling, and then uses positioning to make the chosen construct work. That is the sequence inverted, and it produces two characteristic errors: constructs that were never needed, and constructs chosen too late to have been ordered.

The residual is not a number you report. It is the only thing with standing to change the implant.

What the residual looks like in each joint

The finding this sequence produces

There is a specific and slightly counterintuitive result that falls out of running the sequence properly, and the hip demonstrates it most clearly. When Vigdorchik's group began computing spinopelvic targets rather than defaulting on risk, dual mobility use in their highest-risk group fell from every single case to roughly a third, and the overall dislocation rate held at four point three per cent.

Read that carefully, because the obvious reading is wrong. Better planning did not identify more patients who needed the bailout. It identified the ones who did not. The value of computing a target is at least as much in the escalations it lets you safely decline as in the ones it prompts. That reframes what an instrument like this is for: it is not a device that says use more constraint, it is one that tells you when you have earned the right to use less.

Part II · Seven joints

Each chapter follows the same seven headings so the joints can be read against each other: the envelope, the second look, the fixed reference, what positioning absorbs, the residual and its escalation, the finding, and the honest limit. Each links to a working instrument that computes what the chapter describes.

Chapter fiveThe knee: laxity is a curve, not a number

The knee is where the argument started, because it is where a universal target was abandoned first and most publicly. Kinematic alignment, then functional alignment, moved the field from imposing a neutral axis to restoring a constitutional one and then adjusting components to the soft tissue actually present.

The envelope. Compartment opening under a standardised varus and valgus stress, measured at each of several flexion angles rather than once in extension. This matters because laxity is not constant through the arc: the medial side behaves differently from the lateral, and both behave differently at ninety degrees than at zero. A single extension measurement describes one point on a curve and is routinely treated as the curve.

The second look. Applying the stress at each angle, which requires a deliberate protocol rather than an impression.

The fixed reference. The patient's constitutional alignment, read from their own bone morphology, paired with a physiologic laxity reference. The physiologic reference is itself asymmetric — Matsuda's work puts normal medial opening near two and a half millimetres and the lateral side roughly two and a half degrees laxer than that. A knee balanced to be symmetrical has been balanced to something no normal knee is.

What positioning absorbs. Component coronal angles and femoral rotation, moved inside published functional-alignment boundaries: the femur roughly three degrees varus to six valgus, the tibia two valgus to six varus, rotation three internal to six external, with the final limb axis held within a few degrees of neutral and the joint line preserved. Within that box a great deal of imbalance simply disappears. In Lustig's valgus series, eighty-six per cent of knees landed in the target zone with no soft-tissue release at all.

The residual and its escalation. What the boundaries cannot balance is what should change the implant: sagittal and coronal conformity, lip height, the medial-to-lateral thickness step. A lax compartment gets geometry that supplies stability; a tight one gets geometry that refuses to add constraint.

The finding. Stiffness must be judged against the reference, and per compartment. Averaging the two compartments hides exactly the case that matters — a tight medial side beside a lax lateral one averages to normal and is not normal. The instrument for this chapter got that wrong in its first version and called a stiff varus knee compliant.

Chapter sixThe hip: a cup is aimed once, and the patient stands up

If one chapter has to carry the argument, it is this one, because the hip's universal target failed in a way that is impossible to argue with. Most hips that dislocate were aimed inside the safe zone.

The envelope. Functional pelvic tilt across posture — supine, standing, seated. The pelvis rotates posteriorly when a person sits, and every degree of that rotation adds roughly seven tenths of a degree of functional cup anteversion. A mobile pelvis therefore hands the cup about twenty degrees of protective anteversion for free at precisely the moment the hip is most at risk of posterior dislocation. A stiff one hands it nothing.

The second look. The seated lateral spinopelvic film. It is the single most informative radiograph in this book and it is the one most often not taken.

The fixed reference. Pelvic incidence — fixed morphology that changes with nothing.

What positioning absorbs. Cup anteversion, chosen so that its functional value lands inside an acceptable window in both standing and seated positions rather than in the supine frame where it was measured.

The residual and its escalation. When no single orientation satisfies both postures, positioning has run out and the construct absorbs the difference — a larger head, a lipped liner oriented to the risk direction, or dual mobility.

The finding. Running the four hip-spine groups through the instrument produces a result I did not expect and have not seen stated: deformity alone does not narrow the target window. Immobility does.

The aligned-mobile and flatback-mobile knees have identical windows. A pelvis that still rotates carries its shifted frame into both postures and the geometry never notices. The two stiff groups collapse to four degrees and two and a half. That is the difference between a target you can hit and one you cannot, and it is set by mobility, not by deformity.

Chapter sevenThe shoulder: a budget, and a gate above it

The shoulder introduces something the knee and hip do not have: a soft-tissue finding that overrides the geometry entirely, no matter how favourable the arithmetic underneath it.

The envelope. Not a motion but a displacement — how far the glenoid has eroded from the surface it started as, in version and in posterior subluxation.

The second look. The premorbid glenoid, reconstructed from the vault by three-dimensional planning. Walch's own term for it, the paleoglenoid, tells you the field already thinks of the current surface as something laid over an older one.

The fixed reference. That premorbid surface.

What positioning absorbs. Corrective eccentric reaming, and it has a real ceiling: roughly ten to fifteen degrees before vault perforation and excessive medialisation. In one surgeon survey, seventy-eight per cent cap corrective reaming at fifteen degrees and seventy-three per cent reach for an augment beyond ten. That survey is vendor-published and should be read as a practice pattern rather than evidence, which is how the instrument labels it.

The residual and its escalation. Degrees of correction the reamer should not take are exactly what an augmented baseplate, a graft, or a change of implant class exists to carry.

The finding. The bone budget only decides the case when the rotator cuff works. An incompetent cuff sends the shoulder to a reverse regardless of how correctable the glenoid is. The knee and hip are decided by geometry; the shoulder is decided by geometry underneath a soft-tissue gate, and the instrument shows the gate opening and shutting rather than folding it into the arithmetic. Chapter thirteen returns to this, because once you have seen one gate you start finding others.

Chapter eightThe ankle: the hip's problem, one segment down

The ankle turned out to be the hip again. There is a joint below it that either shares the deformity or refuses to, and a clinical test that tells you which — and the answer changes the operation even when every number on the films is identical.

The envelope. How much of the coronal deformity the hindfoot will give back.

The second look. The Coleman block. Drop the first ray off a block and see whether the hindfoot corrects. It is the ankle's seated film and it costs the price of a piece of wood.

The fixed reference. The physiologic hindfoot — up to about five degrees of valgus is normal and is not deformity to correct.

What positioning absorbs. Intra-articular correction, bounded by what the implant and balancing can take.

The residual and its escalation. What the joint cannot absorb goes below it — a calcaneal or first-ray osteotomy and ligament reconstruction — or, if it cannot go below it, to fusion.

The finding. Two ankles, identical arithmetic, different operations.

Every number the films produce agrees. A tool that stopped at the ankle would call these the same case and be wrong half the time.

Chapter nineThe foot: the chapter that argues against itself

This is the weakest instance in the book and I have kept it in for that reason. Its structure is sound and its second look may not measure what it claims to.

The envelope. First metatarsophalangeal dorsiflexion available under load, against the roughly forty-five to sixty degrees a gait cycle demands.

The second look, and its problem. Passive hallux dorsiflexion in standing — Jack's test, the Hubscher manoeuvre — distinguishing functional from structural limitus. It has been reported not to relate to hallux dorsiflexion during walking. The measurement standing in for what the toe does at push-off may not stand in for it. The hip's seated film and the ankle's block carry no equivalent challenge.

What is nonetheless worth keeping. Two things. The demand framing — asking what the gait cycle requires rather than what the textbook range is — remains useful even if the specific measurement is soft. And the foot has a second two-position test that is not in question: the Silfverskiöld manoeuvre, dorsiflexing the ankle with the knee extended and then flexed, separating an isolated gastrocnemius contracture from a whole triceps one. That matters here because the calf loads the forefoot from above and the toe is downstream of it.

The gate. Pain through the mid-range of passive motion rather than only at its end is the Coughlin and Shurnas grade-four finding, and it overrides the arithmetic completely. It means the surface is gone, and a procedure that preserves motion preserves a painful arc.

Chapter tenThe hand: an adjacent segment with no second position

The thumb shows the adjacent-segment pattern in its clearest form and, at the same time, shows what the method looks like when the kinematic half is missing.

The structure. As the carpometacarpal joint collapses into adduction the thumb cannot get out of the palm, so the metacarpophalangeal joint above it hyperextends to open the web. That compensation is invisible on the radiograph everyone looks at, and beyond roughly thirty degrees the literature treats it as its own decision with options running from doing nothing through capsulodesis and pinning to arthrodesis. Rebuild the base alone and the reconstruction inherits a problem it did not create.

The gate. Clinical, not radiographic. The indication is failure of non-operative management, not the stage on the film — a point the hand literature makes more insistently than most, because stage and symptoms correlate poorly.

The honest limit, which is structural. There is no second position here. The hyperextension is measured once, statically. The adjacent-segment logic is identical to the ankle's and the hip's, but the evidence beneath it is thinner, and the twenty-degree watch threshold in my instrument is my own convention while only the thirty-degree point is published. This is a structural argument wearing the method's clothes, and it should be read that way.

Chapter elevenThe spine: where the universal target is not merely wrong

The spine closes the loop, because it uses the same films as the hip read in the opposite direction, and because it is the one place where achieving the universal target is itself the mechanism of harm.

The relationship to chapter six. The hip asks: will this pelvis move? The spine asks: how much of that motion am I about to remove? A stiff spine is the hip's hazard and the spine operation's product. Same radiographs, same parameters, opposite direction, and to my knowledge the two literatures barely speak.

The fixed reference. The patient's age.

The finding. Take one set of radiographs — pelvic incidence minus lordosis of forty-three degrees, sagittal axis a hundred and fifteen millimetres, pelvic tilt thirty-four, marked on all three modifiers. Now ask what correction they demand.

A twenty-seven degree swing on date of birth alone, from films that are pixel-identical. And the asymmetry is the whole point: the neutral target is wrong in both directions but dangerous in only one. In the older patient it overshoots by nearly seventeen degrees, in an age group where junctional kyphosis already runs near fifty per cent and where the patients who developed it had been overcorrected relative to age-adjusted goals.

Correcting a seventy-eight-year-old spine to a thirty-four-year-old's numbers is not thoroughness. It is the documented mechanism of the failure.

Part III · What follows

Chapter twelveThe adjacent segment

Three times in seven joints, the thing that decided the operation was not the joint being operated on. It was the one next to it, and whether that neighbour would move.

The pelvis for the hip. The subtalar joint for the ankle. The metacarpophalangeal joint for the thumb. Three regions with essentially no shared literature, arriving at the same structure: a primary joint whose planning is dominated by the behaviour of an adjacent one.

The three cases are not identical, and the differences are instructive. The pelvis compensates helpfully and its failure is silent — nothing about a stiff pelvis announces itself in the hip. The subtalar joint compensates passively and its state is revealed by a deliberate test. The thumb's MCP compensates actively and visibly, and is nonetheless routinely not measured, because the radiograph everyone orders does not show it.

The general rule, if there is one

Something like: for any joint, ask what the neighbouring joint is currently doing to accommodate the disease, and what happens to that accommodation after your operation. A neighbour that is compensating and will continue to is protective. A neighbour that is compensating and will be stiffened by your construct is a liability. A neighbour that is already rigid is a constraint that must be planned around before the primary joint is touched.

The test, run after this chapter was written

I originally ended this chapter by stating the rule as a hypothesis and naming the elbow and the wrist as the obvious places to test it. Both have since been built. The result was not what I expected, and it is better than what I expected.

The elbow confirms the pattern more strongly than any joint in this book. The forearm is a ring — the joint at the elbow, the interosseous membrane between, the joint at the wrist. Excise the radial head while the central band is incompetent and the radius migrates proximally. The elbow is fine. The patient presents months later with ulnar-sided wrist pain from impaction, at a joint nobody operated on, often to a different surgeon. That is the adjacent segment at its most extreme: the neighbour is a whole limb segment away, and the failure is so displaced in space and time that it is routinely not attributed to the operation that caused it.

The wrist does not show the pattern at all. No neighbouring joint compensates for radiocarpal collapse. There is no second position that reveals anything. The midcarpal joint does not absorb the disease — in a proximal row carpectomy it is resected. What decides the reconstruction is which articular surfaces are still able to bear load: the operation makes the capitate articulate with the lunate fossa, so both surfaces must be intact, and when the proximal capitate is degenerate the procedure is not harder but excluded.

A hypothesis that survives every test it is given was not given a real one.

What the failure forced: two families, not one

The wrist belongs to the same structural family as the shoulder. Both are decided by a reference built from surviving or premorbid anatomy rather than by a neighbour's behaviour. Once that is visible, the seven original joints sort cleanly into two groups, and the sorting principle is mechanical.

So the corrected rule is narrower and more useful than the one I started with. Ask what the neighbouring joint is doing only where the joints are mechanically coupled — where they form a chain, a linkage or a ring. Where the failure mode is local surface attrition, there is no neighbour to interrogate and the question to ask instead is what is left to build on. Applying the first question to an attrition joint produces nothing, which is exactly what happened when I built the wrist expecting an envelope and found a classifier with a gate.

I am more confident in the refined rule than I was in the original, precisely because a joint was tested that did not fit and the model changed rather than the joint being explained away.

Two further corrections, forced by the girdle and the neck

First: the families are properties of the question, not of the joint. I wrote the table above as though joints belonged to one family or the other. They do not. The shoulder is an attrition problem when you ask about the glenoid surface — how far it has eroded from its premorbid shape — and a coupled problem when you ask about the girdle, where the scapulothoracic articulation is a textbook adjacent segment. Same anatomy, two questions, two families. What sorts into families is the question being asked, and any joint can be asked both.

That correction matters practically, because it means the right move on meeting a new joint is not "which family is this in" but "which question am I asking, and does that question have a neighbour in it".

Second, and larger: the adjacent segment is not always an input. Every instance up to this point treated the neighbour as something you measure beforehand — will the pelvis rotate, will the hindfoot give the deformity back, does the membrane hold. The cervical spine is the case where the neighbour is an output. Fusing a level loads the levels next to it, and symptomatic adjacent-segment disease follows at a rate somebody has actually measured: 2.9 per cent per year, roughly constant, with about a quarter of patients affected within ten years.

It is the only place in this book where the neighbour's failure has a price list.

That completes a loop running through three chapters. The hip page asks whether the spine is stiff and plans a cup around the answer — the neighbour as a hazard. The lumbar chapter is about the operation that produces the stiffness. The cervical chapter puts a rate on the neighbour failing because you fused. Hazard as input, hazard being created, hazard priced. I did not design that sequence; it appeared when the joints were built in the order the reader asked for them.

It also puts the cervical spine and the elbow into a class of their own, which I would now call displaced failure: joints where the consequence of the operation appears somewhere you did not operate. The elbow's radial head excision surfaces as ulnar-sided wrist pain. The cervical fusion surfaces at the level above. In both, the displacement in anatomy and in time is precisely why the failure is so often not attributed to the operation that caused it — and why a planning instrument that looks only at the joint in front of it will never see either coming.

Where age enters, twice, for unrelated reasons

One last thing the neck exposed. Chapter eleven showed the lumbar spine indexing its target to age. The cervical spine does not — a T1 slope of forty-two degrees demands the same lordosis whether the patient is forty-five or seventy-five. What changes with age here is the exposure: how many years the patient spends living with the neighbour you manufactured. Identical radiographs at those two ages produce roughly seventy-six and twenty-four per cent projected cumulative adjacent-segment disease, and that figure is not a target moving, it is a clock running.

So age matters twice in the spine through two mechanisms that have nothing to do with each other, and a surgeon reasoning carefully about one has no particular reason to have considered the other. That is exactly the kind of thing this cross-field exercise is for, and exactly the kind of thing neither subspecialty literature would surface on its own.

Chapter thirteenThe gate above the geometry

In two of the seven joints, an arithmetic-independent finding overrides everything the geometry says. Recognising this class of finding as a class — rather than as a peculiarity of one subspecialty — is one of the things that only becomes visible when you look across fields.

The shoulder's gate is rotator cuff competence. An incompetent cuff sends the case to a reverse regardless of how correctable the glenoid is. The foot's gate is mid-range pain, the Coughlin and Shurnas grade-four finding: pain through the middle of the arc rather than only at its end means the surface is gone, and preserving motion preserves a painful arc. The hand's gate is different in kind but behaves identically — the indication is failure of non-operative management, not the radiographic stage, and a stage III film in a comfortable hand is not an operation.

What these share is that they sit above the measurement, not inside it. They cannot be traded off against a favourable number. A shoulder with a perfect correction budget and no cuff is not a partially good candidate for an anatomic reconstruction; it is not a candidate.

A gate is not a weighted factor. It is a question asked before the arithmetic is allowed to matter.

Why this is worth naming

Because the instinct when building any planning instrument is to fold every input into one score, and scores hide gates. A model that weighted cuff status alongside glenoid version would produce a moderately favourable number for a cuff-deficient shoulder with good bone — and that number would be actively misleading. The three instruments here that have gates keep them structurally separate and show them opening and shutting, because that is what they do clinically.

It also suggests a question worth asking of every joint in the book: what is the finding here that no amount of favourable geometry can overcome? For the knee, plausibly, extension opening at the magnitude of a released collateral. For the spine, plausibly, bone quality that will not hold the construct. I have not modelled either as a gate, and both probably deserve to be.

Chapter fourteenThe ledger, and why it must never rank

Everything so far produces a prediction: this envelope, this residual, this tier. A prediction that is never checked is an opinion with a number attached. The last structural piece is the record that closes the loop.

The proposal is unglamorous. Log what was predicted before the case. Log what was measured during it. Keep the difference. Over enough cases the difference is not noise — it is a description of how one particular clinician's eye reads a particular kind of joint, and it can be used to calibrate their prediction rather than to grade them.

Four properties it must have

1. It calibrates, it does not grade. The output is a correction applied to that clinician's future predictions, not a position on a list.
2. It never ranks. A ledger that ranks becomes a performance instrument, and a performance instrument that clinicians can influence stops recording what actually happened. This is not a policy preference; it is the mechanism by which such records destroy their own data.
3. It has a floor before it speaks. A handful of cases is an anecdote. Somewhere around forty episodes is where a per-clinician residual starts to mean something, and below that the honest output is "not yet".
4. It belongs to the clinician. The calibration record is a description of their judgement. It should travel with them and not be the property of a hospital, a registry, or the vendor of whichever robot happened to be in the room.

That last property is the one with consequences. Most of the data generated by modern surgical technology accrues to the platform rather than the operator. A record of how accurately a particular surgeon predicts a particular joint's behaviour is one of the few things in this whole apparatus that is unambiguously theirs, and it is worth being deliberate about where it lives before the question is settled by default.

Part IV · Honest edges

Chapter fifteenWhat is anchored and what I assumed

Every model in this book mixes published findings with numbers I chose so the model would run. Presenting those two categories in the same typeface would be the single most misleading thing I could do, so here they are separated.

Anchored in published work

Assumptions I made so the models would run

These are not findings. They are the numbers that had to exist for an instrument to produce output, and they are the first things a reader should argue with.

• Conformity slopes in the knee — how much dish curvature to add per millimetre of measured laxity. Invented. Directionally defensible, quantitatively unsupported.
• The coronal step gain — half a millimetre of thickness per millimetre of imbalance, capped. Invented.
• Cup target windows in the hip — the acceptable functional anteversion range standing and seated. The direction is certain; the boundaries are a tuning surface.
• Intra-articular capacity in the ankle — twelve degrees. A planning assumption, not a measured implant property, and named on the page as the number most worth arguing with.
• Cheilectomy and arc-rotation gains in the foot — twenty and fifteen degrees. Assumptions.
• The twenty-degree MCP watch threshold — mine. Only thirty degrees is published.
• Age-adjusted spinal brackets — approximate, reproduced to show shape, not the primary regression.
• Additivity — the ankle model adds ankle and hindfoot deformity as though they were independent. Real feet are coupled.
• Linearity — tilt-to-anteversion, gap-per-degree, and the alignment optimiser are all linearised. Joints are not.

Chapter sixteenWhat would falsify this

An argument that cannot be wrong is not an argument. Here is what would take this one apart, roughly in order of how much damage each would do.

1. The convergence is coincidence

The core claim is that five fields made one error and found one exit. The strongest counter is that they did not — that constitutional alignment, spinopelvic planning, age-adjusted targets and premorbid glenoid reconstruction are four unrelated technical advances that I have grouped by rhetorical similarity rather than by shared structure. What would settle it: a historical analysis of whether these developments share a causal mechanism or merely a description. I have not done that work and it is the weakest joint in the argument.

2. The residual does not predict anything

The method's operational claim is that what positioning cannot absorb is the right trigger for escalation. That is testable and has not been tested. What would falsify it: a cohort in which computed residual fails to predict outcome, or predicts it no better than the surgeon's unaided judgement. If residual is not prognostic, Part One is an elegant description of nothing.

3. The second look is not worth its cost

Every instrument demands an extra observation. If seated films, block tests and vault reconstructions change management in a small enough fraction of cases, the method is true but not worth doing. What would falsify it: a change-in-management rate low enough that the added imaging fails a cost-effectiveness threshold. Note the spine escapes this entirely — its second look is the date of birth.

4. The adjacent-segment pattern is three coincidences — TESTED, AND PARTLY TRUE

This one has been run since the chapter was written, and it is the only condition here with a result. The prediction: build joints with obvious neighbours and see whether the neighbour dominates. The result: the elbow confirmed it more strongly than any joint in the book — the forearm ring displaces the failure of an elbow operation all the way to the wrist. The wrist refuted it — no neighbour compensates, and the operation is decided by which articular surfaces survive.

What that changed: the rule was too broad. It now applies only where joints are mechanically coupled in a chain, a linkage or a ring, and a second family — attrition joints, where the reference is surviving anatomy — was added alongside it. Chapter twelve carries the corrected version. The original claim as written in this book's first version was wrong, and it is wrong in a way I could not have seen without building a joint that disagreed with it.

5. Age-indexing does not generalise

Chapter three raises the possibility that if the right sagittal target is age-dependent, the right cup position and the right knee laxity might be too. What would falsify it: cohorts showing optimal cup or knee targets are stable across decades. That would confine age-indexing to the spine and remove one of the book's more interesting suggestions.

6. My instruments disagree with the surgeons who use them

The most likely and most useful failure. If experienced surgeons run these tools on their own cases and the tiers routinely disagree with what they did and would do again, the models are wrong in ways I cannot see from here. This is the test I most want run, and none of these instruments has yet been used on a real patient by anyone.

A closing note about what this is worth

I have built seven instruments and written the argument that connects them, and not one of them has been used to plan an operation. The distance between a thesis and a validated method is enormous, and I do not want the polish of these tools to disguise how early this is. What is offered here is a way of organising a question that several fields are already answering separately, plus seven worked examples you can put numbers into and argue with.

If the pattern is real, it should be obvious to a surgeon who reads two of these chapters and recognises their own field in the other one. If it is not real, that same surgeon will see the seam immediately. Either outcome is more useful to me than the current state, which is an argument with no readers.

Back matterSources

Cited by claim rather than by chapter, and with the strength of each stated where it matters. This is not a systematic review and does not pretend to be one; it is the set of work these instruments were built against.

1. The hip safe zone fails. Abdel MP et al. What safe zone? The vast majority of dislocated THAs are within the Lewinnek safe zone for acetabular component position. Clin Orthop Relat Res 2016. — PMC4709312
2. Pelvic tilt becomes functional anteversion. Lembeck B et al. Pelvic tilt makes acetabular cup navigation inaccurate. Acta Orthop 2005. — Acta Orthop
3. Hip-spine classification. Vigdorchik JM et al. 2021 Otto Aufranc Award: a simple hip-spine classification for total hip arthroplasty. Bone Joint J 2021;103-B(7). — Bone & Joint
4. Planning replaces the bailout. Decreased instability in high-risk (hip–spine 2B) patients. J Arthroplasty 2024. — PubMed 38642849
5. Functional alignment in valgus knees. Lustig S et al. Knee Surg Sports Traumatol Arthrosc 2025. — KSSTA
6. Physiologic knee laxity is asymmetric. Ligament balancing in TKA — the medial stabilising technique. — PMC5730662
7. Distal resection buys extension. Management of fixed flexion contracture in primary TKA. SICOT-J 2024. — SICOT-J
8. And costs mid-flexion laxity. Cross MB et al. Recutting the distal femur to increase maximal knee extension causes coronal plane laxity in mid-flexion. Knee 2012;19:875–9.
9. Posterior osteophytes and extension. Leie MA et al. J Orthop 2020;19:76–83. — PubMed 32021042
10. The lateral femoral condyle is not hypoplastic. Shah R, Vail T, Bini S. J Orthop Exp Innov 2023 (6,829 MRIs). — JOEI. Corroborated by Howgate et al., J Arthroplasty 2025.
11. Walch glenoid classification. — PMC5539019; Bercik MJ et al., JSES 2016 (3D modification).
12. Anatomic versus reverse for the biconcave glenoid. — PMC10426621
13. Ankle coronal deformity and failure. Influence of preoperative tibiotalar alignment in the coronal plane. Foot Ankle Int 2019. — FAI. Contested by Bone Joint J 2020.
14. Hindfoot alignment matters to TAR survival. — PMC10612447
15. Hallux rigidus grading and the mid-range finding. — PMC5434342
16. And the challenge to the standing test. Weight-bearing passive dorsiflexion of the hallux in standing is not related to hallux dorsiflexion during walking. — via Jack Test references
17. Thumb MCP hyperextension beyond 30°. Management of the metacarpophalangeal joint in thumb CMC arthritis. Hand Clin. — Hand Clinics
18. SRS-Schwab classification. — PMC6259802; Schwab F et al., Neurosurgery 2012.
19. Age-adjusted goals and junctional kyphosis. Lafage R et al. Age-adjusted alignment goals have the potential to reduce PJK. — PubMed 28263226
20. Overcorrection and junctional kyphosis. Spine J 2021. — Spine Journal
21. The episode context. CMS Transforming Episode Accountability Model. — CMS

Back matterColophon and standing

Author. Blaine Warkentine, MD. An MD with deep clinical training who left an orthopaedic residency in year four and has spent twenty years in orthopaedic technology and health system partnerships since — including building a surgical navigation company's orthopaedic vertical to roughly $250M, and as a named inventor on image-guided navigation patents. Not a practising surgeon and not a licensed physician. Nothing in this book is clinical advice.

On the patents. Two of the navigation patents this method descends from — the envelope-point fit and the force-versus-forceless correction vector, both filed in 2008 — are expired and abandoned respectively, which makes them public property that anyone may practise. Several others naming me as an inventor remain active and are owned by my former employer. Inventorship is a credential, not a licence, and nothing in this book builds on the ones I do not own.

The instruments. Seven, each computing what its chapter describes: knee (and its geometry studio), hip, shoulder, ankle, foot, hand, elbow, wrist, girdle, cervical, spine. All are design instruments. None is a cleared device.

How this was made. The models are small numerical programs; the pages are hand-written. Where a search surfaced evidence against something I had built — the lateral condyle, the standing hallux test — the correction is in the text rather than in a revision history, because a thesis that quietly edits out its errors is worth less than one that shows them.