Most discussions of navigation in shoulder arthroplasty start in the wrong place: accuracy numbers. A degree or two of improved version on average is not, by itself, a reason to change your practice. The average glenoid is not the problem. The problem is the glenoid you cannot see properly once you are in the joint, with posterior bone loss, a retroverted face, and a reaming plane that stops looking obvious the moment the cartilage is gone.
I spent six months during my shoulder fellowship in Sydney working with computer-assisted live navigation for exactly these cases, and presented the work on its accuracy in severe glenoid deformity at the ISKSAA–BESS meeting and at SICOT in 2019. What follows is what I took from it — less about the technology, more about where it earns its place.
Where the eye fails
The reliable surgical landmarks for glenoid orientation are surprisingly few. The floor of the supraspinatus fossa and the scapular body give you a sense of version, but once retroversion exceeds about 15–20°, intra-operative judgement of the corrected plane becomes unreliable. Studies of freehand baseplate placement consistently show wide scatter in these deformed glenoids, and the errors are not random — they trend towards leaving the component retroverted, because the eroded posterior bone pulls your reaming with it.
This matters because baseplate malposition is not cosmetic. Excess retroversion and superior tilt are associated with early loosening, scapular notching in reverse arthroplasty, and instability. The cases where you most want to get it right are precisely the cases where the unaided eye is least trustworthy.
What navigation adds, and what it doesn’t
Navigation does not make an easy case better. In a straightforward A1 glenoid with intact bone stock, an experienced shoulder surgeon will place a good baseplate without it, and the registration time is overhead you don’t need.
What it adds is a stable reference frame that survives bone loss. The system registers to scapular anatomy that is not eroded, then reports version, inclination, and seating in real time as you ream and seat the baseplate. In the deformed glenoid, that turns an estimate into a measurement. The published series bear this out: the gain in accuracy is small in normal glenoids and substantial in the B2, B3, and severely retroverted cases.
The honest framing, then, is not “navigation makes you more accurate.” It is navigation narrows the spread of outcomes in your hardest cases, and tells you when you are about to perforate or under-correct. That is a different and more useful claim.
The cost side
Navigation is not free. It adds setup time, capital and per-case cost, and a registration step that can fail and must be repeated. It also introduces a subtle risk: trusting the screen over the bone. The number on the display is only as good as the registration that produced it, and a confident, wrong readout is more dangerous than an honest estimate. The discipline is to use it to confirm what the anatomy is telling you, not to replace looking.
Patient-specific instrumentation and preoperative 3D planning solve part of the same problem more cheaply, and for many glenoids that is enough. Navigation’s distinct advantage is that it is live — it adapts when the bone does not match the plan, which in revision and severe deformity it frequently does not.
How I think about it now
I reach for navigation, or at minimum detailed 3D planning, when the preoperative CT shows retroversion beyond roughly 15°, significant posterior erosion, or a revision where landmarks are already disturbed. For the routine primary, planning suffices. The technology is a tool for the tail of the distribution, not the middle — and that is exactly where it should be.
The interactive companion to this argument — a baseplate you can tilt yourself, with a modelled failure risk — is in Glenoid version, in your hands.