What's Actually Going On
The talus is the linking bone of the hindfoot. It has no muscular attachments, roughly two thirds of its surface is joint cartilage, and its blood arrives backwards through a network entering at the neck, the sinus tarsi and the deltoid branches.
Displacement strips that supply. This is why the risk of the bone dying tracks the severity of the break rather than anything else: 9.8% at the mildest grade rising to 53.4% at the third, with an independent review reproducing the same slope from 0% to 55.0%.
And the bone is buried. It sits between the shin bone above and the heel bone below, with the bony bumps to either side. A break that has not moved produces swelling, weight-bearing pain and tenderness that are, on the surface, a lateral ankle sprain. Every case in the retrieved literature arrived that way.
The deeper problem is a measurement one. Every study validating the Ottawa Ankle Rules used plain radiography as its reference standard. A fracture the X-ray does not show is a fracture the rule's own yardstick scored as absent. The reported sensitivities, whether 99.4% or 91%, are measured against a test that cannot see the thing this page is about. That is not a flaw in those studies. It is a boundary on what they can tell you.
How to Identify It
- Ottawa Ankle Rules Sn: 99.4% / 0.91 | Sp: 35.3% / 0.25
Two pooled estimates, and the spread is the point. Bony tenderness at the posterior edge or tip of either malleolus, at the navicular, or at the base of the fifth metatarsal, plus inability to bear weight four steps. Measured against plain radiography only.
- Deep palpation of the lateral talar neck Sn: NO DATA | Sp: NO DATA
Palpate deep and anterolateral, just distal to the anterior joint line, with the ankle in slight plantarflexion to bring the neck forward. Named as the examination to apply at late review.
- Subtalar glide and side-to-side stress Sn: NO DATA | Sp: NO DATA
Stabilise the talus, move the heel bone. Pain on this movement with a comfortable main ankle joint points away from a simple ligament sprain.
The two NO DATA rows are the honest state of this field, not an omission. No published accuracy figure exists for any physical examination test for this condition, in adults or children. A plausible invented number in a diagnostic accuracy table is the most dangerous thing this page could contain, because it looks exactly like a real one.
The discriminating feature is the trajectory, not any single test. A lateral ankle sprain gets better. This plateaus.
The Debate
How good is the ankle checklist, really?
Bachmann 2003, BMJ, 27 studies pooled, 15,581 patients
Negative likelihood ratio 0.08 for both ankle and midfoot. Sensitivity of almost 100%. The rule is an accurate instrument for excluding fracture.
vs
Gomes 2022, BMC Musculoskelet Disord, 15 studies, 8,560 adults
Sensitivity 0.91 (0.89 to 0.92), specificity 0.25, heterogeneity 94.3%. Roughly one fracture in eleven is not flagged.
The tell sits inside the more flattering paper. Beckenkamp 2017 reports 99.4% across 66 studies, then reports that studies at low risk of bias, and studies where every patient received a radiograph, produced lower accuracy estimates. The best-designed studies give the least flattering answer, which is the usual direction, and Gomes' restricted population sits closer to that subset. Follow the conservative estimate.
Would a different rule fix it?
Barelds 2017, J Emerg Med, 18 studies, 6 rules
Ottawa rules negative likelihood ratio 0.12, against 0.14, 0.23 and 0.39 for the alternatives. The Ottawa rules are the best of the six.
vs
The structural problem
All six were validated against radiography. A better rule scored against the same blind yardstick inherits the same blind spot.
Not a real conflict, and that is what makes it useful. No better rule exists, so switching rules is not the answer. Keep the rule and change what a negative result licenses you to conclude.
No clinical practice guideline was identified for this condition as of 30 August 2026, from NICE, APTA, BOA, EULAR, ACR or JOSPT, and no Cochrane review. This is an absent guideline rather than an out-of-date one, and it is the single largest weakness in the evidence behind this page.
Honest Limitations
1. Every miss-rate number comes from a denominator never built to produce one
The research finding: one series reports 7 missed talar body or neck fractures out of 102 cases. Another opens with a 39% figure.
The real-world gap: the 102 is a selected pool of reviewed cases, so 7 of 102 is not a 7% miss rate. The 39% refers to ankle and midfoot fractures generally being missed at initial emergency evaluation, not to the talus. Either would read as a population incidence to anyone skimming.
Clinical adjustment: carry the mechanism and the consequence, drop the frequency claim. You do not need a prevalence figure to justify palpating a talar neck.
2. The evidence points both ways at once, and both directions are true
The research finding: specificity of 25% to 35%.
The real-world gap: that means the dominant real-world error in ankle triage is over-imaging, not under-imaging. This page describes a small, high-consequence tail, and a clinician who responds by imaging every persistent sprain will be wrong most of the time.
Clinical adjustment: the trigger is three findings together, not one. Persistent pain, plus focal deep bony tenderness at a named talar landmark, plus documented failure to progress. One finding on its own is a re-assessment in two weeks.
3. The outcome literature cannot separate "missed" from "severe"
The research finding: impaired function in 64.7% of non-operatively managed posterior process fractures against 33.3% after surgery, in a group where 36.4% were diagnosed late.
The real-world gap: nothing was randomised. The late-diagnosed fractures and the non-operatively managed fractures are largely the same patients. Delay may be causing harm, or severity may be causing both.
Clinical adjustment: state the association as an association. The case for prompt diagnosis rests on the consequence profile of the fracture itself, which is well quantified, and needs no causal claim this literature cannot support.
4. The imaging you need is the imaging you cannot order
The research finding: MRI made the diagnosis in 7 of 8 cases.
The real-world gap: most physical therapists cannot order an MRI and often cannot order a CT either. The referral succeeds or fails on the quality of the written justification.
Clinical adjustment: the deliverable is a referral letter naming the original X-ray, the exact site of tenderness, and the objective failure to progress. That is a physical therapy finding produced by serial examination nobody else in the pathway performs, and it is what converts a rejected imaging request into an accepted one.
The Nuance
Surgery versus conservative treatment, once the fracture is found. In posterior process fractures, 64.7% of non-operatively managed patients had impaired function against 33.3% after open reduction, and 73.7% of all complications landed in the non-operative group. Across talar fractures generally, pooled success was 62% over 987 fractures with a mean follow-up of about four years, and 60% for neck fractures alone.
The honest truth is that nobody randomised anyone, and no such comparison exists. The non-operative groups in these reviews contain the late diagnoses, the fractures nobody thought worth fixing, and the patients who declined surgery. All three confounders push the result in the same direction, and the apparent superiority of surgery is inflated by an unknown amount as a result.
The finding that survives that problem best is the useful one: across displaced neck fractures, poor outcomes correlated with poor reduction and not with the fracture grade, the surgical approach or the fixation strategy. What is done matters less than how well it ends up sitting.
And the number to quote a patient is not the median. In 36 lateral process fractures followed a mean of 5.5 years, the median hindfoot score was 75 with a range of 12 to 100. The median flatters what individual outcomes in this condition can look like.
Conservative management is not off the table, but the window is narrow and stated narrowly: nondisplaced, small-fragment, extra-articular fractures only.
