Epidemiology and the operational question

Acute traumatic tendon injuries of the hand and wrist occur at an incidence of 33.2 per 100,000 person-years in population-based data, with extensor injuries marginally more common than flexor injuries and with peak incidence in men aged 20 to 291. Zone II of the index finger is the most frequent site of flexor injury. Work-related causes account for around a quarter of cases, predominantly in construction and food preparation. The often-quoted "twenty per cent of hand injuries involve flexor tendon damage" is not directly supported by the primary epidemiology and is best treated as a rule-of-thumb rather than a citable figure.
The operational question for the consultant is rarely whether to repair — almost all clean lacerations are now repaired primarily — but how. The evidence supporting modern practice is dispersed across half a century of biomechanical, anatomical, and clinical work, and the integration of that evidence into a single reproducible pathway is recent. The remainder of this article describes that pathway and the relevant primary literature, taking a position where the evidence supports one and acknowledging where it does not.

Anatomy: zones, pulleys, vincula

Each finger carries two long flexor tendons. The flexor digitorum profundus (FDP) inserts on the distal phalanx and is the sole flexor of the distal interphalangeal joint. The flexor digitorum superficialis (FDS) divides into two slips that decussate around the FDP and insert on the volar middle phalanx, flexing the proximal interphalangeal joint. The thumb's long flexor, the flexor pollicis longus (FPL), is the sole flexor of the thumb interphalangeal joint and is the analogue of the FDP for thumb flexion.
These tendons run within a fibro-osseous tunnel that is anatomically sequential rather than uniform. The tunnel is divided into five annular pulleys (A1 to A5) and three cruciate pulleys (C1 to C3), with the second and fourth annular pulleys being the largest, thickest, and most consistent across cadaveric series2. The A2 and A4 pulleys are the conventionally named "essential" pulleys, in the sense that their loss without compensatory measures produces clinical ; the more recent literature qualifies this — partial venting of A2 and complete venting of A4 do not produce clinical bowstringing in the published series — but the canonical role of A2 and A4 in tendon biomechanics remains intact34.
Within the synovial portion of the sheath, the two tendons receive segmental blood supply from the digital arteries through the — fine mesenteric folds typically organised into a vinculum breve (short, near the insertion) and a vinculum longum (long, more proximal) for each tendon, although the precise vincular anatomy varies considerably between digits and individuals. Diffusion of synovial fluid through the sheath is a second route of nutrition, and chicken and rabbit models from the 1980s established that flexor tendon healing has substantial intrinsic capacity, independent of extrinsic cell sources or vincular flow56. The clinical implication is that adhesions are a complication of healing rather than a necessary part of it — a conceptual point that underwrites all subsequent rehabilitation strategy.
The five Verdan zones are unchanged in their boundaries — Zone I distal to the FDS insertion, Zone II from the proximal A1 pulley to the FDS insertion, Zone III from the distal carpal tunnel to the proximal A1 pulley, Zone IV within the carpal tunnel, and Zone V proximal to it — but Zone II has been usefully subdivided. Tang's IIa, IIb, IIc, and IId sub-zones correspond, respectively, to the FDS insertion, the vincular portion between the insertion and the distal edge of A2, the segment beneath A2, and the proximal portion at A1 and proximal to A274. The subdivision has practical consequences for pulley venting and FDS slip handling, addressed below.

Mechanism and presentation

Open lacerations to the volar surface of the digit account for most flexor tendon injuries; the wound geometry depends on what the digit was doing at the moment of injury, with a flexed posture exposing the tendons distally and an extended posture exposing them proximally. Closed avulsions are less common but distinctive: a sudden forced extension of an actively flexed digit — paradigmatically a hand grasping a jersey during an evasive movement in rugby, gridiron football, or hand-to-hand combat — can avulse the FDP from its distal phalangeal insertion. The ring finger is preferentially involved, as documented in the original Leddy and Packer series of thirty-six cases8. Crush injuries and degenerative ruptures (in rheumatoid arthritis, after distal radius non-union, around carpal cysts) are infrequent but each requires a different operative approach.
Clinical examination follows from the anatomy. With both FDP and FDS lacerated, the finger rests in extension and the cascade — the regular flexion gradient of the resting hand — is broken. The tenodesis effect is the most useful single test: passive wrist extension normally produces passive finger flexion through the long flexors, and a finger that fails to flex with the wrist extended has a discontinuous tendon proximal to the level of the injury. Isolated FDP testing — the proximal interphalangeal joint held in extension by the examiner while the patient is asked to flex the distal joint — and isolated FDS testing — the other digits held in full extension to suppress the common-belly contribution of FDP, with the patient asked to flex the proximal interphalangeal joint of the affected digit — refine the localisation. Digital nerve and vascular examination is non-optional, given the frequency of associated injuries on the radial and ulnar sides of the digit.

Imaging and decision-making

Plain radiographs are taken in the assessment of any volar laceration where a foreign body may be retained or where avulsion fracture is plausible. In jersey finger, the absence of a bony fragment on radiograph implies a soft-tissue avulsion (Leddy and Packer Type I) and increases the urgency of repair within seven days, since the loss of vincular blood supply with proximal retraction places the stump at risk of necrosis8. A bony fragment trapped at A4 indicates a Type III avulsion; a fragment that has separated from the tendon, with the tendon retracting further proximally, the Type IV variant added to the original classification by Smith in 1981. Ultrasound has a confirmatory role for cases in which the level of retraction or the presence of a partial laceration is not clinically apparent, and is particularly useful in postoperative follow-up to detect gap formation. MRI is reserved for chronic and complex cases, including pre-operative planning of two-stage reconstruction.

Timing of repair

Primary repair within 24 to 48 hours of clean injury was the original Verdan paradigm and remains the optimal timing where wound and patient permit. Delayed primary repair, performed within two to three weeks of injury, produces outcomes broadly equivalent to immediate repair in published series, provided the tendon ends can still be retrieved without grafting and the synovial sheath is preserved74. Beyond three weeks, proximal retraction with myostatic shortening and sheath collapse increasingly preclude direct repair; secondary reconstruction through tendon grafting or two-stage Hunter-rod reconstruction becomes the relevant operative choice9.
The Type I jersey finger merits separate consideration. Both vincula are disrupted by the proximal retraction, and the tendon stump is no longer vascularised through them. The original Leddy and Packer series and subsequent experience support repair within seven days, on grounds of stump viability rather than scar maturation8. Type II and Type III avulsions, with the tendon retained at the level of the proximal interphalangeal joint or with a bony fragment caught at A4, retain at least the long vinculum and tolerate a longer interval to repair, though most centres still aim for repair within two weeks.
The wound itself imposes its own constraints. Heavy contamination, gross soft-tissue loss, or a destroyed pulley system favour delayed primary repair after wound stabilisation, with pulley reconstruction either staged or deferred to two-stage reconstruction. The decision is rarely all-or-nothing; the operative judgement is a calibration between the optimal biology — early — and the achievable surgical conditions, which are sometimes later.

Anaesthesia: the case for WALANT

Wide-awake local anaesthesia with no tourniquet — — is now well-supported as the default anaesthetic for elective primary flexor tendon repair, both in the original Higgins series of 122 tendons in 102 patients and in the broader adoption of the technique over the subsequent fifteen years1011. The case for WALANT is partly anatomical and partly conceptual.
The anatomical advantage is intraoperative testing. With the patient awake and the digit unanaesthetised proximally, the surgeon can ask the patient to actively flex through the repair before closure. Higgins and colleagues coined this intraoperative total active movement examination — — and reported that 7 of 122 tendons demonstrated gapping or instability under active flexion at the time of repair; each was identified, re-repaired, and none subsequently ruptured10. The same series found a tendon rupture rate of 3.3% in WALANT primary repairs, with rupture confined to patients who failed protocol adherence postoperatively. The conceptual advantage is a real-time confirmation that the construct under cyclical active load behaves as the surgeon intended — a test that is unavailable under general or regional block.
A 2024 systematic review and meta-analysis of five comparative studies and 624 fingers reported no significant difference between WALANT and general or regional anaesthesia for tendon rupture (odds ratio 1.027, 95% CI 0.450 to 2.342), adhesion or tenolysis rate (odds ratio 0.601), or reoperation rate (odds ratio 1.193), and a small but statistically significant favouring of WALANT for postoperative range of motion (odds ratio 1.641, 95% CI 1.010 to 2.669)12. The interpretation is conservative: WALANT is at least non-inferior, with a modest signal toward better functional outcome and a substantial logistical advantage in not requiring a tourniquet, sedation, or, in many centres, a formal anaesthetic team.
The case against WALANT is narrower than the case for it. Multi-digit injury, pre-existing local anaesthetic intolerance, paediatric patients, and patients with anxiety severe enough to prevent intraoperative cooperation are reasonable indications for general or regional anaesthesia. In all other elective primary repairs, the default position has shifted from "WALANT is an option" to "WALANT is the default unless there is a specific reason for general anaesthesia." This is the position the contemporary literature supports.

Suture technique

The history of flexor tendon suture technique is the history of progressively higher-strand constructs accommodating progressively more aggressive rehabilitation. Kessler's two-strand grasping suture, introduced in 1973, was the dominant primary technique for the better part of two decades, and the modified Kessler with locking loops at each end remained widely taught into the 2000s13. The shift to multistrand techniques was driven by Savage's 1985 in vitro study of a six-strand construct that demonstrated approximately threefold the tensile strength and one-tenth the gap formation under load when compared with the established two-strand techniques tested in parallel14. This work, supplemented by Lim and Tsai's four-strand cruciate repair in the mid-1990s and Sandow and McMahon's Adelaide single-cross-grasp four-strand repair, established the multistrand era15.
Tang's progressive refinement, formalised in the technique and described in detail in the 2007 zone II paper and the 2018 review, brings the multistrand approach into a coherent operative pathway: a four- or six-strand core suture with adequate purchase length (0.7 to 1 cm in each tendon end), modest tension calibrated to slight bunching at the repair site, locking grasps rather than simple grasps where possible, and a peripheral epitendinous mattress to complete the repair74. The 2018 paper made explicit the position that "more is better" — a controlled bulkiness at the repair site is acceptable and indeed beneficial for resistance to gap formation under cyclical loading — overturning the earlier preference for the most parsimonious construct.
The peripheral suture is not a finishing flourish. Wade, Wetherell, and Amis demonstrated in 1989 that the Halsted horizontal-mattress modification of the peripheral repair increased the maximum strength of the construct by 89%, the load to visible gap formation by 93%, and the load to a 2-mm gap by 77% — gains attributable to the technique itself rather than to the suture material16. A 2017 porcine biomechanical study expanded the comparison across six peripheral techniques and found that the peripheral suture adds up to a tenfold increase in stability compared with core suture alone, with Halsted-mattress and Silfverskiöld cross-stitch consistently outperforming simple running stitches17. The peripheral suture is therefore a structural component of the repair rather than a fraying-control adjunct; the choice of peripheral pattern is part of the operative decision rather than its closure.
A 2026 systematic review and meta-analysis of fourteen biomechanical studies addressed the more granular question of locking versus grasping loops within otherwise equivalent core constructs and found a small but significant advantage for locking loops in both ultimate load (mean difference 7.9 N, 95% CI 5.6 to 10) and 2-mm gap load (5.8 N, 3.6 to 8.1), with high inter-study heterogeneity that limits the clinical interpretability of the effect size18. The practical implication is modest: when other variables are held constant, locking loops are biomechanically preferable to simple grasps, but the difference between a four-strand locking-loop construct and a four-strand grasping construct is unlikely to be the determinant of clinical outcome.
A 2023 single-centre randomised trial of fifty patients comparing four-strand and six-strand repairs in zone II reported no significant difference in adhesion or range of motion between the constructs, with two ruptures across both arms; the trial is underpowered for an equivalence claim, with twenty-five patients per arm, but is consistent with the broader pattern that within the multistrand-and-active-motion paradigm, the precise number of strands above four is not the dominant determinant of outcome19. The relevant question is not four versus six strands, but multistrand with peripheral repair and active motion versus the older two-strand with passive motion paradigm. The answer to that question is unambiguous in the contemporary literature.

Pulley management

The operative reality of zone II repair is that a tensioned, slightly bulky multistrand repair must glide through a fibro-osseous canal that was sized for an unrepaired tendon. The classical doctrine that A2 and A4 pulleys must be preserved at all costs has been progressively qualified by twenty-five years of clinical experience, beginning with Kwai Ben and Elliot's 1998 report of 126 zone II lacerations distal to A2, of which 64% required pulley venting of one form or another — most commonly partial venting of A4 (in 56%, with vents of 10% to 100% of the pulley's length) and distal-edge venting of A2 (in 8%, with vents of 4 to 10 millimetres) — without any reported clinical bowstringing or functional deficit3. Tang's 2018 synthesis converges on the same position from independent experience: A2 may be partially vented at its proximal or distal edge, A4 may be vented in its entirety, and complete release of A2 remains contraindicated4.
The operative decision is sub-zone-specific. A repair distal to A2 — Tang's IIb or IIc territory — typically requires venting of A4 and sometimes the distal edge of A2 to permit gliding of the bulky multistrand construct. A repair proximal to A2 — IId — generally tolerates A1 release without bowstringing. A repair beneath A2 — IIc — is the situation in which partial A2 venting is most often required, and the operative compromise is most often a millimetre-scale edge release rather than transverse division. Pulley reconstruction at the time of primary repair is rarely necessary if these venting limits are observed; it is reserved for two-stage reconstruction or for revision of pre-existing pulley failure.

Zone I avulsion and the Leddy and Packer typology

The original Leddy and Packer paper described three types of FDP avulsion in their 1977 series of thirty-six athletes, classified by the proximal extent of tendon retraction and the presence or absence of a bony fragment at the insertion8. In Type I, the tendon retracts to the palm with both vincula disrupted and no bony fragment; Type I demands repair within seven days because the avascular stump degenerates rapidly. In Type II, the tendon retracts only as far as the proximal interphalangeal joint, where it is held by the long vinculum, often with a small bony fragment; the long vinculum preserves a partial blood supply and tolerates a longer interval to repair. In Type III, a larger bony fragment lodges at A4 and prevents proximal retraction beyond the middle phalanx; the fragment may be amenable to fixation as well as tendon repair. The Type IV variant — a fragment from which the tendon has dissociated and retracted further — was added by Smith in 1981, and the Type V variant, with a concurrent distal phalanx fracture, by Al-Qattan in 2001. The original three types and these later additions remain the working framework for jersey finger management.
The operative choice between bone-anchor reattachment, transosseous pull-out suture, and dorsal-button techniques is not strongly differentiated by the available evidence; small comparative series suggest equivalent outcomes. The dominant determinants of outcome in jersey finger are timing — within seven days for Type I, within two weeks for Types II and III — and recognition that the FDP, alone among finger flexors, must transmit force across an anchored end-fixation rather than a tendon-to-tendon suture line. The biomechanical demands are different, and so are the rehabilitation thresholds; a bone-anchor reattachment will not tolerate the same active flexion forces as a multistrand tendon-to-tendon repair, and modern protocols qualify active mobilisation accordingly.

Rehabilitation

The history of flexor tendon rehabilitation has tracked the history of suture technique. The Duran controlled-passive-motion protocol, published as a book chapter in 1975, was designed for the strength of a two-strand Kessler construct and prescribed passive flexion exercises within a dorsal blocking splint, with no active flexion until six weeks. The Kleinert dynamic protocol, formalised in 1981, introduced rubber-band traction to provide passive flexion against active extension and remained widely used through the 1990s. Both protocols accepted a substantial trade-off — high adhesion and decreased range-of-motion rates in exchange for low rerupture rates — that reflected the strength limit of the available repair.
The Belfast regimen described by Small, Brennen, and Colville in 1989 was the first widely cited early-active-motion protocol for zone II repair, with active flexion within 48 hours and good or excellent outcomes in 77% of 138 zone II injuries20. The dehiscence rate of 9.4% in that early-active-motion series was historically high but was a consequence of two-strand repair under active load, not a refutation of the active-motion concept itself. Subsequent active-motion regimens — Indianapolis place-and-hold, the Saint John and Manchester variants, the Nottingham regimen — operate within the same conceptual framework: active flexion within the first week or two of repair, with the construct strength of the underlying repair determining the magnitude and frequency of permitted force.
The pivotal randomised comparison was Trumble and colleagues' 2010 trial of 103 patients comparing place-and-hold active flexion with passive motion, with 52-week follow-up in 93 patients and 106 digits21. The primary outcome — interphalangeal joint motion at final follow-up — was 156° in the active group and 128° in the passive group, a clinically and statistically significant difference favouring active mobilisation. Secondary outcomes — flexion contracture, satisfaction — also favoured the active group. Tendon ruptures were equal in number (two each) and not statistically different. The Trumble trial established active mobilisation as biomechanically and functionally preferable when the underlying repair is sufficiently strong to tolerate it.
Starr and colleagues' 2013 systematic review of thirty-four articles addressed the broader pooled comparison and reported rerupture rates of 4% — 57 of 1,598 tendon repairs — for early passive motion and 5% — 75 of 1,412 — for early active motion, with the trade-off in decreased range of motion (9% in the passive group, 6% in the active group) going the opposite way22. The interpretation is that early passive motion has a marginally lower rerupture rate but a higher decreased-ROM rate, while early active motion inverts this trade-off; the practical balance has shifted toward active motion because the modern multistrand construct makes the rerupture difference negligible while preserving the ROM advantage.
A 2022 zone-II-specific meta-analysis of seven studies and 569 digits confirmed this position: the active-motion advantage in range of motion is preserved with multistrand repair, while the rerupture risk that previously argued against active motion is now confined to two-strand constructs23. A 2024 systematic review of twenty-eight studies and 1,414 patients reached the same conclusion across a longer span24. The contemporary patient-reported-outcome data, in a 2025 prospective cohort of thirty-two patients with sixty-one repairs followed under early active mobilisation, reported total active motion of 83.5% of the unaffected hand at three months, return to work in 100%, and return to leisure activities in 96.6%25.
The position the evidence supports is that early active mobilisation, calibrated to the strength of the underlying repair and to the tendon-to-bone or tendon-to-tendon nature of the repair, is the standard of care for primary zone II flexor tendon repair under multistrand fixation. Passive-only protocols remain appropriate where the construct is intrinsically less strong — bone-anchor reattachment in jersey finger being the canonical example.

Contemporary outcomes

The aggregate complication burden of flexor tendon repair, as reported in Dy and colleagues' 2012 meta-analysis of 138 articles and 4,145 repairs, is a 6% reoperation rate, a 4% rupture rate, and a 4% adhesion rate26. The same meta-analysis reported that the use of an epitendinous suture decreased the reoperation rate by 84% and that modified Kessler core sutures decreased the adhesion rate by 57% relative to other core techniques pooled together. A sobering finding from the same paper was that complication rates did not differ significantly between repairs performed before and after the year 2000 — the temporal improvement that one would expect from the introduction of multistrand techniques and active-motion rehabilitation was not visible in the pooled meta-analytic data through 2012.
The contemporary best-practice benchmark is more encouraging. Pan and colleagues' 2019 series of 60 fingers repaired with a tensioned strong four- or six-strand core, sparse peripheral stitches, judicious A2 and A4 venting, and early active motion reported zero ruptures, 87% good or excellent outcomes by Tang criteria, and a follow-up range of 8 to 33 months27. The combination — multistrand, peripheral, venting, active motion — produces outcomes that are categorically different from those that pooled twentieth-century data report. Higgins and colleagues' WALANT series of 122 tendons in 102 patients reported a 3.3% rupture rate, with all ruptures attributable to protocol non-adherence rather than to construct failure; the 7 of 122 tendons that demonstrated intraoperative gapping under active testing and were re-repaired did not subsequently rupture10.
The pooled rerupture rate from the older systematic reviews — 4 to 5% in the Starr 2013 synthesis — reflects the era of two-strand repair and passive-only or early-passive mobilisation22. The contemporary rerupture rate, when multistrand repair, peripheral suture, judicious venting, intraoperative active testing, and early active rehabilitation are combined, is in the 0–4% range and is typically attributable to protocol non-adherence rather than to construct failure. The variation in published outcomes between centres is increasingly explicable not by underlying patient or injury factors but by the degree to which centres have adopted the integrated paradigm. Outcome benchmarking that does not control for which paradigm is in use produces misleadingly pessimistic figures.

Two-stage reconstruction

When primary repair has failed, when the wound is too contaminated for primary repair, or when the patient presents beyond the window for direct repair with retraction and sheath collapse, the operative choice is two-stage reconstruction along the lines of the Hunter and Salisbury 1971 protocol9. The first stage places a silicone gliding implant — the Hunter rod — within the reconstructed sheath, with pulley reconstruction as required, and is followed by twelve weeks of passive motion to develop a synovial pseudosheath around the implant. The second stage replaces the implant with an autologous tendon graft — palmaris longus most commonly, plantaris or a sacrificed FDS as alternatives — and proceeds to a graded rehabilitation protocol that mirrors primary repair, allowing for the higher fragility of the tendon-to-tendon junctions at the proximal and distal motor units.
Contemporary outcomes are categorically inferior to those of primary repair. Strickland-Glogovac good or excellent results in published series typically range from 50% to 70%, and the overall complication burden — graft rupture, adhesions, joint stiffness, pulley failure — is substantially higher.

Complications and how to think about them

Adhesions are the most prevalent complication of flexor tendon repair, with the operative implication that the early rehabilitation protocol must achieve sufficient tendon excursion to break up nascent fibrous tethering before it matures. Where adhesion has matured to the point of clinically restricting active motion despite full passive motion at six months, — surgical division of the peritendinous adhesions — is the appropriate operative response. The underlying repair must be intact and the patient cooperative with subsequent therapy; tenolysis on a non-compliant patient or on a partially failed repair will not improve function.
Rerupture in the modern paradigm is uncommon and is most often a consequence of protocol non-adherence rather than primary construct failure, as the Higgins WALANT series made explicit10. When rupture is detected, re-repair should be performed within a week, before the construct ends become unrecoverable; the Belfast experience suggests that re-repair followed by the same active-motion protocol can yield good or excellent results in the majority of cases20.
Joint stiffness, particularly proximal interphalangeal joint flexion contracture, is a complication of the rehabilitation protocol as much as of the repair itself. Bowstringing, when it occurs, follows from violation of the venting limits — partial release of A2 beyond the millimetre-scale edge release, or repeated venting at adjacent levels — rather than from a single judicious decision at the time of primary repair. Infection in primary acute repair is rare (under one to two per cent in published series) and is more often associated with delayed presentation than with operative technique.

Future directions and synthesis

The active research frontier in flexor tendon repair operates at the level of biology rather than mechanics. The largest body of preclinical work concerns reduction of post-repair adhesions through modulation of the synovial environment — exogenous lubricin, hyaluronic acid, surface modification of the tendon, and various growth-factor and cell-based strategies have all shown promise in animal models, but none have yet produced level-one human evidence sufficient to translate into routine practice. The clinical trials that would establish these adjuncts as standard remain to be done, and the prudent position for the current generation of consultants is to follow the literature without yet adopting the interventions.
The mechanical dimension of the technique is closer to a stable equilibrium than to active development. Multistrand core suture, peripheral epitendinous repair, judicious pulley venting, and early active mobilisation under wide-awake anaesthesia constitute a paradigm that the published outcome data validate at the level of zone II repair. Marginal questions persist — locking versus grasping loops, four versus six strands, the optimal peripheral pattern, the precise dose of early active motion — but the gross architecture of the operation is settled. The remaining heterogeneity in published outcomes between centres is largely attributable to the unevenness of paradigm adoption rather than to genuine disagreement about technique.
The position the evidence supports for primary repair of zone II flexor tendon injury is therefore as follows. Wide-awake local anaesthesia with intraoperative active testing is the default. A four- or six-strand core suture with locking grasps and a peripheral mattress is the construct. Partial venting of A2 within a millimetre-scale edge threshold and complete venting of A4 are tolerated where the multistrand repair requires gliding clearance. Early active mobilisation, calibrated to the strength of the underlying repair, begins within the first week. The expected rerupture rate is 0–4%, the expected good-or-excellent functional outcome rate exceeds 80%, and the centre-level variation in outcomes now reflects the integration of these elements rather than disagreement about any of them. Where the integrated paradigm is followed, the operation is no longer one of the most challenging tasks in hand surgery; it is one of the most reproducibly successful.