The scaphoid is the most frequently fractured carpal bone, and despite more than a century of clinical attention it remains one of the most consequential injuries in upper-limb trauma. Its anatomy, retrograde blood supply and biomechanical role at the link between the proximal and distal carpal rows mean that small clinical decisions — to image or not, to cast or fix, to graft or salvage — translate over years into preserved or lost wrist function. The publication of the Scaphoid Waist Internal Fixation for Fractures Trial (SWIFFT) in 20201, its five-year follow-up in 20262, the EFORT European guideline review3, the British Society for Surgery of the Hand standards4 and NICE NG385 have together produced a more coherent evidence base than has existed for a generation. This article integrates that evidence with current European practice. It is intended for hand surgeons making day-to-day decisions about acute scaphoid injuries, established non-unions, and salvage of post-traumatic carpal collapse.

Epidemiology and demographics

Population-based data from a defined adult Edinburgh cohort reported an annual incidence of 29 per 100,000, with a clear male predominance and a younger median age in men6. UK general-population data from Leicester reported a lower incidence of 12.4 per 100,000 per year, with the highest incidence in men aged 15–19 and a steep socioeconomic gradient7. Earlier Scandinavian work from Bergen reported 4.3 per 100,000 with 82% in men and a mean age of 258. The figure most often quoted in textbooks — that the scaphoid accounts for 70–80% of carpal fractures — derives from older institutional series and over-represents the bone's contribution; modern defined-population studies favour 60–70%6,7. The clinically important point is the demographic profile rather than the precise denominator: a young, predominantly male, sport- or work-injured population, in which a missed fracture has decades to declare itself.
The Edinburgh cohort distribution by Herbert subtype showed B2 (complete waist) injuries in approximately one-third6. The classical proportional distribution — waist around 65%, proximal third around 25%, distal third around 10% — varies by series and by classification system used, but remains a reasonable working estimate.

Functional anatomy and vascular supply

The is approximately 80% covered by articular cartilage, leaving a narrow non-articular dorsal ridge through which the principal vascular pedicle enters. The canonical vascular study remains Gelberman and Menon's 1980 cadaveric injection-clearing work, which demonstrated that 70–80% of intra-osseous vascularity, including the entire proximal pole, derives from dorsal ridge branches of the radial artery, while the remaining 20–30% — supplying the distal tubercle — arises from volar radial branches9. The intraosseous flow is therefore predominantly retrograde from distal to proximal: a fracture across the waist or proximal third tends to devascularise the proximal fragment, and explains both the time-to-union gradient observed clinically and the disproportionate burden of avascular necrosis () in proximal pole injuries. Gelberman and Menon also concluded that a volar operative approach is least traumatic to the proximal-pole supply, a principle that still guides waist fracture surgery.
The dorsal arterial pedicle has been used both as an anatomical caution — avoiding the dorsal ridge during dorsal mini-open or percutaneous antegrade fixation — and as a donor source for the pedicled bone graft, originally described by Zaidemberg and colleagues in 199110.

Mechanism, classification and the assessment of displacement

The classical mechanism is a fall on an outstretched, hyperextended and radially deviated wrist, with the scaphoid loaded against the dorsal lip of the radius. In sport, axial loads through a clenched fist (boxing, martial arts, snowboard falls) produce comparable patterns. Several classifications coexist; the most widely used in European practice is the Herbert and Herbert–Fisher classification, which incorporates location, stability and chronicity (Type A: stable acute; B: unstable acute, including B1 distal oblique, B2 displaced waist, B3 proximal pole, B4 trans-scaphoid perilunate, B5 comminuted; C: delayed union; D: established non-union)11. The Mayo classification subdivides anatomical region (proximal pole, waist, distal third, distal tubercle), while Russe classified by fracture line orientation. For routine reporting the Herbert system remains the European default.
The pivotal threshold for treatment decisions is displacement. SWIFFT randomised adults with bicortical waist fractures displaced by ≤ 2 mm; this is the operationalised definition of "minimally displaced" in current European practice1. The Clementson, Björkman and Thomsen review draws a finer distinction — non- or minimally displaced (≤ 0.5 mm), moderately displaced (0.5–1.5 mm), and displaced (≥ 1.5 mm) — and recommends internal fixation for all waist fractures displaced ≥ 1.5 mm, prolonged immobilisation (8–10 weeks) for moderate displacement, and standard six-week casting for minimally displaced patterns3. The two thresholds (SWIFFT's 2 mm and Clementson's 1.5 mm) are not in conflict: SWIFFT confirmed cast non-inferiority up to 2 mm in the trial population, Clementson advocates earlier intervention based on a broader cohort literature. Pragmatically, BSSH/GIRFT 2024 standards adopt the 2 mm threshold for primary surgical fixation in waist fractures4.
Displacement assessment requires CT, ideally with reformatting in the long axis of the scaphoid (sagittal-oblique and coronal-oblique planes)3. Plain radiographs systematically underestimate displacement. High-resolution peripheral quantitative CT (HR-pQCT) achieves an interobserver kappa of approximately 0.91 for scaphoid fracture detection but availability is limited to research centres12.

Clinical assessment

Anatomical , scaphoid tubercle tenderness, and pain on axial compression of the thumb are the traditional clinical triad. Used in isolation each is sensitive but non-specific; in combination, and at sequential examinations, their predictive value rises sharply. Duckworth and colleagues prospectively examined 223 patients with suspected scaphoid injury and identified four independent predictors of true fracture: male sex, sports mechanism, snuffbox pain on ulnar deviation within 72 hours, and tubercle tenderness at two weeks13. With all four positive, the post-test probability of fracture rose to 91%; conversely, absence of snuffbox pain on ulnar deviation within 72 hours had a perfect negative predictive value in their cohort. These data underpin the contemporary European emphasis on a structured second clinical review at 10–14 days rather than indefinite empirical immobilisation.

Imaging

A four-view scaphoid series — posteroanterior, lateral, ulnar-deviated PA, and 45° semi-pronated oblique — remains first-line, and BSSH standards stipulate this view set4. When initial radiographs are negative but suspicion persists, the contemporary European pathway has departed from older "repeat-radiograph at two weeks" protocols. NICE NG38 recommends MRI as first-line imaging for suspected scaphoid fracture after thorough clinical examination, on cost-effectiveness grounds — early definitive diagnosis avoids prolonged unnecessary immobilisation and the medico-legal exposure of missed fractures5. The Cochrane review by Mallee and colleagues reported pooled summary sensitivity and specificity of 0.88 and 1.00 for MRI versus 0.72 and 0.99 for CT in radiographically occult scaphoid fracture, with bone scintigraphy more sensitive but considerably less specific — the over-treatment rate of scintigraphy makes it an inferior gatekeeper for a low-prevalence presentation14. Yin and colleagues' meta-analysis with latent class analysis found pooled MRI sensitivity 97.7% and specificity 99.8% versus CT sensitivity 85.2% and specificity 99.5%15.
The prospective head-to-head data of Sahu and colleagues are useful at consultant level. In 100 patients with clinically suspected occult scaphoid fracture and negative radiographs and CT, MRI demonstrated additional fractures in 16% (15 patients), of whom 8% had a scaphoid fracture and 10% had bone bruising; 25% of patients required hand-surgery follow-up16. The clear inference is that MRI captures fractures that CT misses, and that CT — although excellent for displacement, fracture morphology and assessment of union — is not an adequate occult-fracture screen.
CT remains the imaging of choice for assessing displacement, and union once a fracture is known. Scans should be reformatted in the long axis of the scaphoid; conventional wrist-plane CT systematically underestimates the morphological detail required for surgical planning3. Photon-counting CT and dual-energy CT are emerging modalities; specific high-quality scaphoid evidence remains limited.

Acute management of waist fractures: the SWIFFT evidence

For two decades a steady drift towards primary surgical fixation of minimally displaced waist fractures was driven by retrospective evidence of faster return to work, with mixed long-term data. SWIFFT was the methodologically definitive answer. In a pragmatic, multicentre, open-label superiority trial across 31 hospitals in England and Wales, 439 adults — mean age 33, 83% male — with a bicortical scaphoid waist fracture displaced by ≤ 2 mm were randomised to immediate surgical fixation or to below-elbow cast immobilisation, with surgical fixation reserved in the cast arm for confirmed non-union1. At 52 weeks the adjusted difference was −2.1 (95% CI −5.8 to 1.6; p=0.27) — no clinically meaningful difference. Surgical complications were higher (14% versus 1%) and cast-related complications were higher in the cast arm (18% versus 2%). The trial's pre-specified conclusion was that adult patients with scaphoid waist fractures displaced by 2 mm or less should have initial cast immobilisation, with any suspected non-union confirmed and immediately fixed1.
The five-year follow-up published in 2026 collected valid PRWE scores from 344 patients (78.4% of the original cohort). The mean PRWE difference between groups was 0.6 (95% CI −2.4 to 3.6; p=0.709) — clinical equivalence sustained at five years2. Among patients with five-year imaging, only 3 of 146 (2.1%) in the fixation group and 4 of 121 (3.3%) in the immobilisation group had non-union. Grip strength and range of motion were comparable. The accompanying radiological analysis demonstrated that fractures with at least 20% bridging at one year went on to consolidate without intervention, and that osteoarthritic progression was similar in both groups (52.4% had osteoarthritis in at least one peri-scaphoid joint at five years), with worse PRWE scores correlating with greater osteoarthritis severity17. The cost-effectiveness analysis found initial cast immobilisation to be the more cost-effective strategy, with a mean lifetime saving of approximately £1,606 per eligible patient18. The earlier 1-year economic evaluation had reported a probability of cost-effectiveness for early surgery of 5.6% at the UK NICE threshold19.
The contemporary recommendation, congruent with the EFORT review and the BSSH standards, is therefore: minimally displaced (≤ 2 mm) waist fractures should be treated initially in cast, with early surgical fixation reserved for confirmed delayed or non-union; this is also the most cost-effective pathway over five years. The argument for primary fixation now rests on patient-specific factors — return-to-work or return-to-sport pressures, occupational requirements, hand dominance — rather than on superior long-term outcome.
The most recent meta-analysis pooling 12 RCTs concluded that surgery shortens healing time and accelerates return to work and grip recovery without affecting non-union rate or long-term function20. SWIFFT remains the largest contributing trial.

Cast type and duration

The historical default of a thumb-spica long-arm cast has been progressively dismantled. Buijze and colleagues randomised 62 patients with confirmed non- or minimally displaced waist fractures to a below-elbow cast including or excluding the thumb; the cast-excluding-thumb group demonstrated significantly greater bridging at 10 weeks (85% versus 70%), with overall union of 98%21. The earlier meta-analysis of randomised trials by Doornberg and colleagues had already concluded that no nonoperative immobilisation method — above- versus below-elbow, thumb-included versus excluded, neutral versus flexed/extended wrist — produced superior union rates22. The 2022 systematic review of Siotos and colleagues confirmed equivalent non-union rates for short-arm versus long-arm cast (9.6% versus 10.5%) and for thumb-included versus thumb-free cast (10.3% versus 10.1%)23. The combined evidence supports a below-elbow cast, in neutral wrist position, with the thumb free as the contemporary default for non- and minimally displaced waist fractures, typically for six weeks with CT-confirmed assessment of union thereafter — or eight to ten weeks for the moderately displaced 0.5–1.5 mm group.
The follow-up imaging strategy matters as much as the immobilisation choice. Plain radiographs at six weeks are unreliable for confirming union and routinely overestimate healing. CT in the long axis of the scaphoid at six to eight weeks is the appropriate confirmatory study, with bridging trabeculation of at least 50% of the cross-section as a working threshold for cast removal. The SWIFFT five-year radiological data confirm that fractures with as little as 20% bridging at one year went on to consolidate without further intervention, which provides reassurance for ambiguous mid-term imaging17. Patients are counselled that mild radiocarpal stiffness following cast removal is expected and that grip strength typically recovers to within 90% of contralateral by three to four months without formal hand therapy. Persistent snuffbox tenderness or an unimproving range of motion at six weeks following cast removal warrants repeat imaging to exclude delayed union.

Distal pole and proximal pole fractures

Distal pole fractures — including tubercle avulsions and distal third fractures — heal predictably and rapidly. The majority unite uneventfully after four to six weeks of immobilisation; non-union is rare3. Standard management is below-elbow cast immobilisation with a low threshold for early mobilisation once tenderness has resolved.
Proximal pole fractures behave very differently. Eastley, Singh, Dias and Taub's meta-analysis pooled 1,147 acute scaphoid fractures managed nonoperatively, of which 67 (5.8%) were proximal24. Aggregated data showed that 34% of acute proximal scaphoid fractures progress to non-union with conservative management, with a relative risk of non-union of 7.5 compared with more distal fractures. The authors recommend that the proximal scaphoid be defined as the proximal fifth of the bone to avoid misclassification, and that CT be used during follow-up. For the consultant, this evidence supports primary internal fixation for most proximal pole fractures, typically with antegrade dorsal headless compression screw placement; the screw should be central in both AP and lateral views, perpendicular to the fracture, with the start point judiciously placed to avoid the dorsal ridge vascular pedicle. Pragmatically, primary fixation remains the standard in most European centres for proximal pole fractures, with conservative trial reserved for clearly minimally displaced injuries in compliant patients accepting prolonged casting4.
of the proximal pole is reported in 13–40% of proximal-pole fractures, the variation explained by differences in fracture-line definition (proximal fifth versus proximal third), imaging modality (plain radiographic sclerosis versus MRI loss of marrow signal versus intraoperative punctate bleeding), and chronicity at presentation. The risk approaches 100% in fractures isolating the proximal fifth, where the dorsal ridge supply is interrupted9. AVN of the waist or distal pole is uncommon. The operating principle is that any proximal pole fracture or non-union should be assessed preoperatively with MRI for proximal-fragment viability when surgical reconstruction is planned, although MRI signal alone does not perfectly predict intraoperative vascularity, and final assessment is made by direct observation of bleeding from the proximal fragment after debridement.

Surgical fixation: percutaneous, mini-open and arthroscopic

For minimally displaced waist fractures requiring fixation, percutaneous antegrade or retrograde headless compression screw placement is the standard. Central screw placement is biomechanically superior to eccentric placement, and trajectory planning increasingly relies on long-axis CT reformats and, in some centres, patient-specific instrumentation. The screw should occupy the central one-third of the proximal and distal fragments in both the AP and lateral planes, with sufficient working length to engage subchondral bone at both poles without violating the articular surfaces. Screw length is determined by intraoperative measurement minus 4 mm to ensure adequate compression without prominence. Open volar (Russe) or dorsal mini-open approaches are used for displaced waist fractures and for proximal pole fractures respectively, the latter requiring meticulous protection of the dorsal ridge vascular pedicle through a transverse incision over the lateral edge of the third extensor compartment. The Slade arthroscopic-assisted percutaneous technique offers an intermediate option, combining direct visualisation of fracture reduction with the soft-tissue sparing of the percutaneous approach.
Arthroscopic-assisted reduction and fixation has matured into an accepted technique in centres with appropriate equipment and operator experience. Its advantages include direct visualisation of fracture reduction, assessment of associated ligament injury, and minimal soft-tissue dissection. Its limitations are operative time, learning curve and equipment dependence; it has not been shown in randomised data to deliver superior long-term outcomes versus open or fluoroscopy-guided percutaneous fixation, but it is increasingly the preferred technique for displaced waist fractures and for injuries (Herbert B4) in specialist centres. SWIFFT's secondary analysis of associated ligament injury reported scapholunate gap widening of more than 3 mm on initial imaging in only 2.8% of the trial population, with no progression at five years — reassurance that occult ligament injury in the minimally displaced waist fracture population is uncommon and clinically inconsequential.
has been proposed for unstable waist fractures and selected non-unions to provide rotational stability. Biomechanical and clinical series suggest theoretical advantages in resisting torsional loads25, and a small recent series reported 100% union at a mean of 2.7 months in 21 displaced non-unions managed with double-screw fixation and bone grafting26. High-quality comparative evidence remains limited; outside specific situations — oblique fracture pattern, revision fixation, larger scaphoid in male patients with adequate bone stock — single-screw fixation remains standard.

Established non-union: stratified treatment

Scaphoid non-union is conventionally defined as failure of union beyond six months, although modern practice increasingly intervenes earlier when imaging at three to four months demonstrates static fracture configuration without progressive consolidation. Without intervention, established non-union progresses through a predictable pattern of carpal collapse — radial styloid impingement, scapholunate dissociation pattern, and eventually . Mack and colleagues' natural history study, with follow-up of up to 53 years, demonstrated three radiographic stages with fracture displacement and lunate dorsiflexion correlating with severity27. Ruby, Stinson and Belsky's parallel study found that 31 of 32 patients (97%) at five years or more from injury had developed wrist osteoarthritis, with the single exception having concomitant AVN28. Untreated scaphoid non-union progresses to wrist osteoarthritis in essentially all patients given enough time.
Surgical management is determined by location, presence of humpback deformity, viability of the proximal fragment, prior surgery, and patient factors. For non-union without AVN and without significant humpback deformity, non-vascularised cancellous or corticocancellous bone graft from the iliac crest or distal radius, combined with screw or wire fixation, remains highly effective. For non-union with humpback deformity and dorsal intercalated segmental instability (), structural graft (Russe or Fisk-Fernandez wedge) restores scaphoid length and corrects the carpal alignment.
The vascularised versus non-vascularised graft debate has been reframed by recent meta-analyses. Three independent recent systematic reviews now converge on a higher union rate for than in established scaphoid non-union, particularly with proximal-pole AVN and in revision cases29. In simpler cases without AVN or significant deformity the difference is smaller and clinically marginal. The practical implication is that the choice of graft is now a stratified decision rather than a categorical preference. For routine waist non-union without AVN and without significant deformity, non-vascularised distal radius or iliac crest graft is appropriate. For proximal pole non-union with confirmed AVN, vascularised reconstruction is preferred. For humpback deformity, structural graft remains the standard, with vascularised pedicled options available as alternatives. For revision after failed non-vascularised graft, vascularised reconstruction is the default.
The 1,2-ICSRA pedicled graft from the dorsal distal radius, popularised by Zaidemberg and refined by multiple authors, harvests vascularised bone on a retrograde pedicle10. Reported union rates in heterogeneous series are favourable but variable, particularly disappointing in proximal pole AVN, where the pedicle may not deliver adequate inflow against established osteonecrosis. This shortcoming has driven the rise of the free corticoperiosteal flap and the related medial femoral trochlea (MFT) osteo-chondral flap for proximal pole reconstruction, particularly when the proximal articular surface is itself compromised. Free vascularised reconstruction from the femur is technically demanding and requires microsurgical capability; in centres without that infrastructure the sensible referral pathway is to a tertiary hand or hand-and-microsurgery service rather than committing to repeat pedicled graft attempts with diminishing returns.

SNAC wrist: staging and salvage

Untreated scaphoid non-union progresses through Watson's three-stage SNAC pattern: stage I — radial styloid arthrosis, with intact midcarpal joints; stage II — extension to the scaphocapitate articulation; stage III — periscapholunate arthrosis with extension to the capitolunate joint. The lunate–radius articulation is preserved until late, which is the anatomical basis for the most commonly used salvage options. Each stage is identifiable on plain radiographs supplemented by CT for cartilage assessment; MRI is rarely required for staging decisions but contributes when proximal-fragment viability is in question and motion-preserving reconstruction is being considered.
For SNAC stage I, radial styloidectomy combined with scaphoid reconstruction may be considered when the proximal fragment remains viable and the patient retains an essentially preserved midcarpal articulation. The styloidectomy should be limited to 4–6 mm to avoid destabilising the radioscaphocapitate ligament. For stage II and III, the choice is between and . The contemporary evidence broadly favours PRC for older, lower-demand patients with intact lunate fossa cartilage and a viable proximal capitate articular surface, on the grounds of preserved range of motion, simpler technique, and lower hardware-related complications. Four-corner fusion preserves the capitolunate articulation and is preferred when the proximal capitate is affected or in younger high-demand patients accepting greater range-of-motion loss for durability. Total wrist arthrodesis remains the final salvage for failed motion-preserving procedures or pan-carpal arthrosis. The choice between motion-preserving salvage options is appropriately individualised; the literature does not support a single dominant strategy across all SNAC presentations.

Paediatric scaphoid fractures

Paediatric scaphoid fractures merit comment. Historically these were predominantly distal-pole avulsions with rapid healing; the contemporary epidemiology shows a shift towards adult-pattern injury — 71% waist, 23% distal pole, 6% proximal pole, in a cohort of 351 paediatric fractures30. Casting alone delivers 90% union in acute paediatric fractures, but lower union rates obtain with chronic, displaced or proximal injuries; surgery achieves 96.5% union. Almost one-third of paediatric scaphoid fractures present late with established non-union, which should be managed with surgical reduction, fixation and grafting. The implication is that paediatric scaphoid injuries should not be dismissed as benign and that the same imaging principles — low threshold for MRI, axis-aligned CT for displacement assessment — apply.

Return to work and sport

European practice for return to sport in elite or semi-elite athletes increasingly favours percutaneous fixation of even minimally displaced fractures to permit earlier return to training in protective splinting, recognising that this is a patient-preference decision rather than an outcome-superior pathway. Typical return to non-contact training is at four to six weeks following percutaneous fixation, and to contact sport at eight to twelve weeks subject to CT-confirmed union. For cast-treated minimally displaced fractures, return to contact sport is typically at ten to twelve weeks with CT confirmation of union. SWIFFT did not specifically address athletic populations, and the evidence base for return-to-sport timelines remains predominantly observational; consultant-level decision-making in this group is appropriately individualised.
Smoking is consistently associated with delayed and failed union in scaphoid surgery. Active counselling and, where realistic, perioperative smoking cessation should be a routine component of consent for scaphoid non-union surgery. Diabetes, vitamin D deficiency, and chronic glucocorticoid use are additional patient-level factors that should prompt a more conservative threshold for primary fixation in acute injury and for vascularised reconstruction in revision cases.

A practical decision framework

Drawing the evidence together, a contemporary approach to acute scaphoid injury can be framed as follows. Patients presenting with a fall on the outstretched hand and snuffbox pain on ulnar deviation within 72 hours should have a four-view scaphoid radiograph series. If positive for a non- or minimally displaced waist fracture (≤ 2 mm, bicortical), below-elbow cast immobilisation with the thumb free for six weeks is the SWIFFT-supported default, with surgical fixation reserved for confirmed delayed or non-union; this is also the most cost-effective pathway over five years. If the radiographs show a displaced waist fracture — beyond 1.5 to 2 mm step or gap — or humpback deformity, trans-scaphoid perilunate injury, or a proximal pole fracture, primary internal fixation with a headless compression screw (open or arthroscopic-assisted) is recommended. If the radiographs are negative but suspicion remains, MRI within three to five days is the NICE- and EFORT-supported standard, with CT reserved for assessment of displacement and union once a fracture is confirmed.
For non-union, treatment is stratified by location and biology. Waist non-union without humpback or AVN: non-vascularised distal radius or iliac crest graft with screw fixation. Waist non-union with humpback or DISI: structural non-vascularised wedge graft with screw fixation. Proximal pole non-union with AVN: vascularised reconstruction, with free MFC or MFT flap as the most reliable option in tertiary centres; pedicled 1,2-ICSRA may be considered when AVN is mild. Revision after failed non-vascularised graft: vascularised reconstruction. Established SNAC wrist: PRC or four-corner fusion stratified by cartilage status, age and demand.

Outstanding uncertainties

Several areas remain incompletely resolved. The optimal management of acute proximal pole fractures detected on MRI in minimally displaced configuration — primary fixation versus extended cast — has never been the subject of an adequately powered randomised trial; Eastley and colleagues' power calculation suggested 76 cases would be required, and this study has not been performed24. The role of arthroscopic-assisted fixation of waist fractures versus standard percutaneous fixation lacks high-quality comparative data. Long-term outcomes of resurfacing PRC and of MFT flaps are still being accrued. The role of HR-pQCT, photon-counting CT and dual-energy CT in routine pathways remains restricted by availability rather than by diagnostic performance. The literature on adjuncts — low-intensity pulsed ultrasound, bone morphogenetic protein, autologous bone marrow aspirate concentrate — remains low-quality and inconclusive; none of these adjuncts has demonstrated robust randomised superiority over standard graft and fixation.
Scaphoid fracture management has converged on a coherent evidence base over the last decade. The SWIFFT trial and its 2026 five-year follow-up have settled the principal acute question. NICE NG38, the BSSH standards and the EFORT 2020 guideline review have aligned imaging pathways around early MRI and long-axis CT. The Eastley meta-analysis and modern reviews have reset expectations for proximal pole fractures, where primary fixation remains the prudent default. Recent meta-analyses have reframed the vascularised versus non-vascularised graft debate as a stratified rather than categorical decision. The classical Gelberman and Menon vascular anatomy still explains the clinical pattern of failure 45 years after publication, and remains the proper primary citation for blood-supply teaching. For the consultant, the discipline now lies less in adopting a particular technique than in correctly stratifying the fracture or non-union and matching it to the appropriate point on the evidence-graded pathway.