Extensor tendon injuries are among the most commonly encountered upper-limb trauma presentations and among the most easily under-treated. The dorsal apparatus is thin, superficial and biomechanically interdependent; small departures from anatomically and temporally appropriate care produce extensor lag, tendon adhesion or fixed deformity that no later operation reliably reverses. Decision-making centres on the eight-zone classification ratified by the Committee on Tendon Injuries of the International Federation of Societies for Surgery of the Hand1 and elaborated by Doyle in successive editions of Green's Operative Hand Surgery2, with the thumb separately zoned T-I to T-V. The contemporary evidence base — anchored by the long-term outcomes series of Newport3, the systematic syntheses of relative-motion rehabilitation45, and the British Society for Surgery of the Hand trauma standards6 — has clarified what was once a field of competing dogmas into a small set of reasonably defensible defaults. What it has not done is resolve the harder problems: the limits of conservative management, the place of surgical fixation in mallet finger, and the threshold for repair of partial lacerations all remain governed more by expert convention than by level I trial evidence.

Anatomy and the zonal scheme

The extensor mechanism is conventionally divided into eight zones running distal to proximal — odd-numbered zones overlying joints (zone I at the DIP, III at the PIP, V at the MCP, VII at the wrist) and even-numbered zones over the intervening segments. Doyle added a ninth zone for the proximal forearm musculotendinous junction in the textbook elaboration of the original Kleinert-Verdan scheme2. A 2024 letter by Tang7 noted persistent inconsistency in zone numbering across textbook chapters and proposed a unified five-zone reformulation aligned with the flexor scheme; the eight-zone Verdan-Doyle convention nonetheless remains the operative classification in current European training and examination.
Three anatomical features carry disproportionate clinical weight. First, the over the proximal phalanx divides into a inserting on the dorsal base of the middle phalanx and two that reunite distally to form the terminal tendon — disruption of either limb produces a characteristic and self-perpetuating deformity. Second, the interconnect the extensor digitorum communis tendons over the dorsum of the hand, allowing an isolated proximal laceration to be partially compensated by neighbouring tendons and producing the classic missed zone-VI injury. Third, the centralises the extensor tendon over the MCP joint; the radial sagittal band is anatomically thinner and longer than the ulnar component, predisposing it to rupture and producing ulnar tendon subluxation when injured8. These three features — the central-slip / lateral-band division, the juncturae, and the asymmetric sagittal-band anatomy — are the substrate of nearly all the deformities described below.

Diagnosis

Most extensor tendon injuries are diagnosed clinically. Open injuries declare themselves; closed injuries require deliberate examination of each joint in isolation, with the wrist and adjacent MCP joints positioned to neutralise contributions from the juncturae and the long extensors. The 99 for central-slip integrity — the patient extending the PIP against resistance with the joint flexed over a table edge — remains a key bedside manoeuvre.9 Houston and colleagues found in a cadaver model that the change in DIP resistance immediately after central-slip transection could be too small to appreciate clinically and became more detectable after repeated flexion–extension cycles.9 These experiments do not establish a five-to-ten-day clinical interval.9 An equivocal early test warrants continued assessment of the suspected injury and appropriate follow-up; deliberate unprotected cycling is not a diagnostic recommendation.
Plain radiographs are obtained in essentially every significant injury — to identify avulsion fragments at the terminal tendon and central-slip insertions, to detect retained foreign material, and in the setting of suspected fight-bite injury, to demonstrate metacarpal-head chip fractures or air in the joint space. Cross-sectional imaging is reserved for problems clinical examination cannot resolve. Bedside ultrasound with linear high-frequency probe is well-suited to direct visualisation of the dorsal tendons; it is operator-dependent but learns rapidly. The most recent comparative diagnostic data come from a series of sixty late-presenting tendon injuries (median delay months rather than days), in which ultrasound sensitivity for extensor tendon injury was eighty-four per cent and MRI sensitivity was forty-four per cent, with both modalities under-performing the figures reported in earlier acute-presentation cohorts because of established peritendinous scar10. The figures apply to the late-presentation context in which they were derived; comparable head-to-head acute-presentation data are limited.

Mallet finger and zone I injury

The mallet finger — terminal tendon disruption with or without bony avulsion — has accumulated a more elaborate classification system than its management justifies. Two classifications coexist in the literature. The Wehbé and Schneider scheme11, a bony-mallet-only system, distinguishes injury without DIP joint subluxation (Type I), injury with subluxation (Type II) and transphyseal paediatric injury (Type III), with subtypes A, B and C reflecting fragment size of less than one third, one third to two thirds, and more than two thirds of the articular surface respectively. Doyle's parallel scheme2 classifies all mallet injuries: closed (Type I), open with tendon laceration (Type II), open with substance loss (Type III) and bony (Type IV, subdivided by mechanism and articular involvement).2 The single hard indication for surgery is volar subluxation of the distal phalanx — when the joint cannot be reduced congruently in the splint, the biomechanics no longer support tendon-to-bone or fragment-to-bone healing in any reasonable position12. The widely repeated thresholds for fragment size — one third or one half of the articular surface — are softer criteria. They were not findings of the original Wehbé and Schneider series, which concluded that most mallet fractures could be managed conservatively, the size and position of the fragment notwithstanding11; the 2017 review of Lamaris and Matthew12 notes explicitly that an advantage of surgery in complicated cases — those defined either by fragment size or by volar subluxation — "has yet to be clearly proven."
The contemporary evidence is consistent on this. The 2004 Cochrane review13 found insufficient evidence to favour any particular splint type or to define when surgery is indicated, and no Cochrane update has been published since. The 2023 meta-analysis of Peng and colleagues14 — the highest-level synthesis available — found no high-level evidence of superiority of surgery over orthosis for either bony or tendinous mallet injury, with surgical and conservative management producing comparable extensor lag and range of motion. The retrospective comparative series of Nagura and colleagues15 reported a mean DIP extension lag of 2.1° after K-wire fixation versus 13.8° after splinting in fifty-nine patients, but the splint arm contained three patients explicitly recorded as non-compliant, and the cohort is small, retrospective and unblinded. The Aksan single-centre series of one hundred and ninety patients across four management arms16 found mean extension lag of 6° after splinting alone, 6.1° after splint-then-K-wire salvage, 3.8° after primary K-wire and 17.3° in patients who declined treatment — figures that bear out the conventional position more robustly: splinting and pinning yield comparable results in compliant patients, with K-wire fixation reasonable when compliance fails.
The reasonable practical protocol distinguishes the bony from the purely tendinous injury. Bony mallet — fracture-to-fracture healing on a vascularised bone-bone interface — heals on the standard fracture timeline, and six weeks of continuous DIP extension splinting is the durable convention, supported by the Tolkien national survey17 in which seventy-eight per cent of UK clinicians reported a six-week protocol for bony injury. Pure tendinous mallet is the slower problem. Tendon-to-bone scarring at the terminal insertion lacks the biological substrate of fracture healing, and conservative success in this group depends on prolonged immobilisation: 17the British survey reported continuous DIP splinting for eight weeks in fifty-six per cent of respondents and six weeks in thirty-nine per cent17, and a Korean series of one hundred patients managed with twelve weeks full-time stack splinting followed by four weeks of night-only orthosis reported a final mean extensor lag of 2.6° and seventy-eight per cent successful results by Abouna and Brown criteria18. The DIP is splinted in extension while the PIP remains free.19 For a bony mallet injury, the position must maintain joint congruity19 without excessive hyperextension or pressure on the dorsal skin20; an in-splint lateral radiograph and follow-up imaging where indicated assess reduction.19 Persistent volar subluxation of the distal phalanx in the splint — the joint not seated in extension despite splint correction — is the indication to convert to surgery, irrespective of fragment size.
Success in this context is realistic, not absolute. A residual extensor lag of approximately ten degrees should be the expected outcome of conservative management in the compliant patient and discussed at the time of splint application; published series report final extensor lag in successful cases ranging from 2.6° (Hong) to approximately 14° (Nagura splint arm), with most series clustering between five and ten degrees181516. The functional outcome — restored composite finger flexion, absence of nail-bed deformity, patient satisfaction — matters more than the goniometric lag in isolation. from untreated terminal tendon disruption is the classical late complication and motivates timely intervention even when the acute deformity appears trivial.

Central slip and the boutonnière injury

Acute closed disruption of the central slip — Verdan zone III — produces, over one to three weeks, the deformity as the lateral bands migrate volarly to the PIP axis and the unopposed terminal tendon hyperextends the DIP. The clinical priority is therefore not the immediate appearance of the digit but the recognition of central-slip injury before the deformity establishes. Splinting the PIP in full extension while leaving the DIP free permits the disrupted central slip to heal at length and obliges the patient to mobilise the lateral bands dorsally during the period of immobilisation21. Six weeks of continuous splinting9 followed by a period of nighttime wear is a commonly used protocol. The available comparative evidence does not establish that a four-week protocol produces more residual extensor lag than a six-week protocol.22 Relative-motion flexion splinting — holding the affected MCP in slight flexion relative to its neighbours — has emerging evidence as an alternative permitting earlier hand use23, but the comparative trial data remain narrower than for the zone V–VI relative-motion extension protocols discussed below.
Open central-slip lacerations require operative repair followed by the same six-week protective regime. Established boutonnière deformity — present at three months or longer — is a different problem entirely, in which restoration of PIP extension reliably trades one deformity for another by producing a flexion deficit; reconstruction is therefore approached cautiously and with the patient's specific functional priorities defining the goal of surgery rather than a standardised target21.

Sagittal band injury and the closed zone V dislocation

Closed sagittal band injury at the MCP — most often the radial sagittal band of the long or ring finger after a direct blow with the fist clenched — produces ulnar subluxation of the extensor tendon and an inability to initiate MCP extension that often resolves with passive positioning. The Rayan and Murray classification24 distinguishes injury without extensor instability (Type I), injury with tendon subluxation (Type II), and injury with frank tendon dislocation (Type III). Despite intuitive appeal the older textbook formulation of "Type I = contusion" is a misreading of the original paper, which defines all three types as injuries to the band itself, distinguished by the resulting tendon behaviour rather than by the severity of band disruption.
The defining clinical fact about sagittal band injury is that acute presentation responds to conservative management even in Type III dislocation. The Catalano series25 of ten patients with eleven Rayan-Murray Type III injuries managed in a custom splint holding the affected MCP in twenty-five to thirty-five degrees of relative hyperextension produced eight of ten patients pain-free, with seven of eleven injuries showing no or barely discernible residual subluxation. 26The systematic review of Wu and colleagues across four hundred and twenty-nine treated digits supports splinting for acute sagittal band injury, although the underlying studies are low-quality and non-comparative.26 James and colleagues describe treatment within three weeks as generally avoiding surgery; this is not a mandatory operative cutoff.27 Peelman and colleagues also reported successful splintage in subacute and chronic presentations, with lower success as chronicity increased.28 Surgical reconstruction — most often by the technique described by Watson, Weinzweig and Guidera, in which a retrograde slip of the extensor tendon is woven through the deep transverse metacarpal ligament29 — is reserved for chronic injury, for failure of conservative management, and for the small minority of acute injuries with associated open laceration of the band.
The fight-bite injury — a punch to a tooth, producing a small dorsal laceration over the MCP with frequently unrecognised joint penetration — sits at the intersection of zone V tendon injury and contaminated joint sepsis. The immediate priorities are urgent surgical exploration, thorough irrigation and debridement, and prompt broad-spectrum antibiotics selected according to local guidance. The British Society for Surgery of the Hand recommends initial surgery within twenty-four hours, sooner for a fight bite or penetrating joint injury. Tendon repair may be delayed until the wound is free of contamination and devitalised tissue; subsequent timing depends on infection, tissue viability and reconstructive requirements, rather than a fixed five-to-ten-day interval.

Open lacerations and repair: zones V–VIII

For open injuries from zone V proximally, surgical repair within the British Society for Surgery of the Hand four-day window is the established standard6. The flat tendon profile in zones V and VI accommodates a figure-of-eight or modified Becker-style suture configuration; the rounder tendon in zones VII and VIII accepts core suture techniques analogous to flexor tendon repair. The published comparative literature on suture configuration in extensor tendon repair is biomechanical and cadaveric rather than clinical, and the contemporary informed default is to use the strongest configuration the tendon profile allows in a manner that supports early protected motion. Where the extensor retinaculum is divided in zone VII, partial Z-lengthening prevents bowstringing without producing dynamic insufficiency.
The canonical long-term outcomes series remains that of Newport, Blair and Steyers3 — sixty-two patients across one hundred and one digits managed predominantly with static post-operative immobilisation. Sixty-four per cent of patients without associated injury achieved good or excellent outcomes by the Miller criteria; results in distal zones (I–IV) were significantly poorer than in proximal zones (V–VIII); and loss of flexion was a more frequent and more functionally consequential outcome than loss of extension. That last finding has shaped the subsequent thirty-five years of rehabilitation thinking.

Partial lacerations and the fifty-per-cent rule

The convention that extensor tendon lacerations involving more than half the cross-sectional area of the tendon require formal repair while those involving less may be managed conservatively is one of the most widely repeated principles in hand surgery and one of the least formally evidenced. The 2023 systematic review of Dickson and colleagues30 identified five studies (one randomised, four cohort) with a total cohort too heterogeneous for meta-analysis; pinch and grip strength outcomes and time to return to work were similar between repaired and non-repaired partial lacerations, and adverse events occurred only in the repair groups. The authors conclude that there is "limited evidence to support the management of partial extensor tendon lacerations" and that "non-operative management of selected partial lacerations is safe."
The fifty-per-cent threshold has the status of pragmatic convention rather than a trial-derived rule.30 Al-Qattan reported non-operative treatment of forty-five partial extensor tendon lacerations involving 55–90% of tendon width, with no rupture, triggering or extensor lag at final follow-up. The series included finger zones II, IV and VI–VIII but excluded zones I, III and V.31 These selected results do not establish a general distal-versus-proximal repair hierarchy.31 Management should account for tendon function, the precise zone and structures injured, wound and joint involvement, and the rehabilitation plan; relative-motion splinting should not be generalized to all distal partial injuries.

Rehabilitation: relative motion and WALANT

The decisive change in extensor tendon management over the past two decades has been the shift from static post-operative immobilisation to early protected active motion. The original Howell, Merritt and Robinson series of one hundred and forty patients across twenty years of practice32 established the immediate-controlled-active-motion protocol, in which a relative-motion orthosis holds the involved digit in fifteen to twenty degrees of MCP hyperextension relative to its uninjured neighbours. The position transfers extensor force from the repair to the adjacent intact tendons through the juncturae, allowing the patient to mobilise the digit actively from the early post-operative days without straining the repair. The original series reported no tendon ruptures, mean return-to-work in eighteen days, and a mean total programme duration of seven weeks.
The contemporary evidence has caught up with the clinical pattern. The Hirth scoping review33 surveyed the broader application of across hand conditions; the Wong systematic review5 found early-motion protocols superior to static immobilisation for total active motion and grip strength outcomes in zones IV–VIII. The Collocott randomised trial34 — forty-two patients in zones V and VI, the highest-quality randomised data on this question — demonstrated superior hand function and patient-reported outcomes at four weeks for relative-motion extension splinting compared with controlled active motion in a wrist-hand-finger orthosis, with no ruptures in either arm. The Shaw systematic review4 of five hundred and twenty-one patients across zones IV–VII reported eighty-nine per cent good or excellent extension and eighty-seven per cent good or excellent flexion by Miller's criteria, concluding that relative-motion rehabilitation is safe in zones V–VI and that the evidence base in zones IV and VII remains developing. The Bojnec randomised trial35 — fifty patients across three rehabilitation arms — showed that the relative-motion-extension protocol with or without a night wrist orthosis produced superior early hand function compared to dynamic extension orthosis, and that the simpler relative-motion-extension-only protocol was equivalent to the more complex relative-motion-with-night-wrist-orthosis variant.
The intra-operative correlate of early motion is wide-awake local anaesthesia with no tourniquet, in which the patient is asked to actively flex and extend the finger on the operating table to verify repair integrity and absence of tendon gapping before closure. The Merritt, Wong and Lalonde review of contemporary extensor management23 documents the technique's spread and its compatibility with relative-motion rehabilitation; the synthesis between intra-operative active motion testing and immediate post-operative active motion is the single most consequential development in the field of the past decade.

Synthesis

The contemporary management of extensor tendon injury sits between two pressures. The first is institutional — the expectation, encoded in trauma standards and audit cycles6, that significant injury is recognised, referred and operated within defined intervals, with hand-therapist-led structured rehabilitation. The second is intellectual — the recognition, slow in arriving, that several historical doctrines on which surgical practice rests are softer evidence than their universal repetition implies. The fragment-size thresholds for surgical management of bony mallet and the fifty-per-cent threshold for partial laceration repair are pragmatic conventions rather than trial-derived rules; the hard indications are narrower and more biomechanically grounded.
The reasonable contemporary stance is to begin with splinting for most closed mallet injuries. The BSSH protocol uses six weeks for bony mallet and eight weeks for soft-tissue mallet, with the DIP in extension and the PIP free; fracture congruity and skin tolerance must be checked, and persistent subluxation warrants discussion of operative and non-operative options.19 Splinting is also an appropriate first-line option for acute closed sagittal band injuries, including Type III dislocation,25 without treating three weeks as an automatic cutoff for surgery.28 Open lacerations are managed within the BSSH four-day window with figure-of-eight or core-suture configurations matched to tendon profile. Partial lacerations require assessment of tendon function, injury zone, wound and joint involvement, and the rehabilitation plan; the available evidence does not support a universal lower repair threshold in zones VI–VII.3031 Relative-motion-extension rehabilitation is the standard post-operative protocol for zone V–VI repair, with WALANT permitting intra-operative active motion testing before closure. The clinical priority across all of this is the same as it has always been: to recognise the injury before deformity establishes, to choose the management whose risk profile most closely matches the patient's compliance and functional priorities, and to rehabilitate aggressively enough to recover flexion without destabilising the repair.