Hallux rigidus

Last evidence check December 2012

Principal authors: Tim Board, Anna Fox, Jim Barrie

“Hallux rigidus” is a term accredited to Cotterill in 1887 and refers to degenerative disease of the first metatarsophalangeal joint with stiffness and deformity. It is the second commonest problem of the 1 st MTP joint (hallux valgus is first). Other terms in the literature include “hallux limitus” (used by podiatrists to refer to limited movement in the 1st MTPJ), “hallux flexus” (flexion deformity with reduced dorsiflexion) and “dorsal bunion”.

Biomechanics of the 1st MTP joint

These are based on a cadaveric study by Ahn et al. (1997)

The average suface contact area within the joint decreases as the proximal phalanx dorsiflexes on the first metatarsal head. In the neutral position, average surface contact area is 0.38cm2; however at full dorsiflexion it decreases to 0.04cm2. The contact point on the proximal phalanx articular cartilage remains constant, however the contact point on the head of the first metatarsal shifts dorsally with increasing toe extension. This fits with the observed chondral erosions and degenerative changes that affect the dorsal aspect of the joint. In addition implant arthroplasty of this joint tends to fail from dorsal loading.


Predisposing features


HR may be secondary to trauma, gout, infection or other inflammatory arthropathy. Degeneration may begin dorsally with limitation of dorsiflexion and impingement as the predominant features, or begin as a central crater spreading to the whole joint. Joint destruction may be worse on the lateral side producing a “hallux rigidus et valgus”.

Camasta (1996) proposed a series of conclusions on the natural history of the disease based on clinical examination and radiographic findings.

  1. Predisposing features causing spastic contracture of the Hallux.
  2. Axis of movement in the 1 st MTPJ shifts from centrally to plantar.
  3. Dorsal articular cartilage impinges resulting in chronic erosion (chondritis dessecans) or subchondral bone fracture (osteochondritis dessicans).
  4. Progressive arthritis and radiographic features similar to OA.
  5. Synovial effusion and splinting due to pain.
  6. Autofusion resulting in reduced range of movement (note this is primarily seen in Rheumatoid disease).

However, he did not offer any data to support these hypotheses.

In addition the sesamoids are subject to degeneration. Hypertrophy of the sesamoids in due to chronic traction, disuse osteopenia is seen, and degeneration parallels that of the MTP joint.

Natural history

Hallux rigidus is a fairly benign condition. Smith et al (2000) reviewed 22 patients 12-19 years after a diagnosis of hallux rigidus. Although there had been radiographic progression in most patients, few had noticed deteriorating symptoms, and only 25% wished to consider surgery at final review. However, in Pons et al's (2007) RCT comparing steroid and hyaluronate injections, half of their patients had undergone surgery at 1 year. The difference between these two series might reflect differences in initial patient populations, and our experience is much more in line with Smith.

This is important in counselling patients, as they are often concerned about progressive disease and keen to “get something done before it gets worse”. We can reassure them that in most people it does not “get worse” and that treatment can be based on current problems rather on an attempt to stave off future disability.


Pain is the commonest feature. In the early stages the pain is mainly felt dorsally and is provoled by dorsal impingement. Later it is felt generally within the joint and is provoked by movement at any point in the range. Some patients have predominantly plantar pain, probably due to sesamoid-metatarsal OA. The pain may radiate down the hallux or up the first ray, and occasionally is felt mainly at a site distant from the MTPJ, but can be provoked by MTPJ movement.

The next commonest complaint is of pain related to a prominent dorsal osteophyte, especially where it rubs on the shoe. There may be a bursa, or the skin may be subject to abrasion, breakdown or infection. Occasionally a patient presents with an ulcer over the osteophyte which communicates with the MTPJ.

A few patients complain mainly of the stiffness, or of the compensatory hyperextension of the IPJ that often develops as the MTPJ get stiffer. Some patients complain the joint locks – this is probably due to synovial impingement as loose bodies are rare.

Patients tend to walk off a painful first ray and may complain of pain under the lateral metatarsals or down the lateral foot border.

There may also be other features of a systemic condition such as gout or rheumatoid disease.

As usual, it is important to determine the exact problem the patient is complaining of. A patient who is bothered only by the osteophyte may be a candidate for a cheilectomy even if the joint is fairly poor. A patient whose main complaint is stiffness may find it difficult to accept that there is no reliable way to improve this.

The patient’s functional demands also have a major influence on management:



The standard radiographs are weight-bearing AP and lateral views. Look for:

Other imaging techniques are not usually necessary. Occasionally an acute presentation, usually with a history of trauma, may be due to an osteochondral injury of the metatarsal head – an MR can be useful in this setting.


A clinical/radiographic grading system was described by Regnauld and appears mainly in the European literature. Hattrup and Johnson (1988) described a radiographic classification which has become standard, and in fact corelates quite well with the Regnauld grading:

Coughlin et al (2003) modified the Hattrup and Johnson classification. He introduced a grade 0, with stiffness and loss motion, but no pain or radiographic changes. Grade 3 was divided into those with advanced arthritic changes but pain only in dorsiflexion or plantarflexion (new grade 3) and those with pain throughout the range of movement (grade 4). This was useful to guide treatment, as cheilectomy was shown to fail in grade 4, but had a variable outcome in grade 3. In addition, Coughlin made the criteria for grading more complex, which often makes gradings less reproducible. There has been no study of the reproducibility of this or other gradings.



One of the most important non-surgical interventions is advice and explanation of the natural history. Patients are generally relieved to hear that this is normally a benign, non-progressive condition and that they can continue any desired activities, including sports, as they are able without any concern that they are storing up trouble for the future.

Choice of shoes which accommodate the stiff joint and dorsal exostosis, and use of a stiff-soled shoe are also useful, as are simple analgesics. Orthoses can be useful in the group of patients (relatively small in our experience) with overpronation and hypermobile first rays. Scherer et al (2006) showed that an orthosis made from a cast with the first ray plantar-flexed increased the range of 1st MTPJ movement in stance and reduced the pressure under the first MT head in gait. We would not consider further intervention until these simple measures have been tried.

Some patients find steroid injections useful in the short to medium term. Pons et al (2007) reported a single-blind RCT comparing steroid with hyluronan injections. The hyaluronan group had better AOFAS scores but there were no other major differences. The evaluators were not blinded, increasing the risk of bias.

There are few papers on the non-surgical management of hallux rigidus. Grady et al (2002) reported that 56% of 772 patients with hallux rigidus were successfully treated non-operatively, using a combination of orthoses, shoe alterations and steroid injections. Our experience is that a larger proportion of patients require only non-surgical treatment.

Manipulation under anaesthesia

Solan et al (2001) reported the results of MUA and steroid injection in 29 patients. Eight of 12 patients with grade 1 and 6 of 18 with grade 2 radiographs had not required any other surgery, but all 5 with grade 3 changes had undergone definitive surgery. It is impossible to tell whether the MUA or injection, or both, were effective. This seems to be a technique of limited use in patients with early disease.


Most authors use a dorsomedial approach, which gives good access to both sides of the joint for osteophyte excision. Mann, however, reported equivalent results with a true medial approach to avoid the dorsomedial cutaneous nerve, while Lin (2009) used a dorsolateral approach with equivalent results but a 40% incidence of minor numbness in the first web space.

The joint is opened dorsally and a variable amount of the dorsal part of the first MT head excised. Some authors remove only the osteophytes, others up to 50% of the metatarsal head. Mann recommended debridement adequate to obtain 70deg dorsiflexion on the table. One cadaver study (Heller et al 1997) shows that 30% excision as recommended by Mann (1988) increases range of movement in hallux rigidus, although this is achieved by abnormal hinging at the dorsal lip. However the reported clinical results are similar irrespective of the amount of metatarsal head resected, though there are no formal comparison clinical studies.

Patients can mobilise immediately weightbearing. Many surgeons advise early active and passive mobilisation of the toe to maintain the range of movement achieved on table. No studies have compared different post-operative regimes.

There have been a large number of series of cheilectomies reported, amounting for over 300 patients. The largest series and the longest follow-up (10 years) was by Coughlin (2003). Coughlin distinguished between patients with severe radiographic change but no mid-range pain (in whom cheilectomy was successful in 32/34 patients and the mean AOFAS score was 90) and those with severe changes and mid-range pain (in whom only 1/9 patients who chose cheilectomy got a satisfactory result, five were revised to arthrodesis and the mean AOFAS score in the remainder was 70). In another long-term series by Waizy (2009), 20/20 patients in grade 1, 23/35 patients in grade 2 and 4/5 patients in grade 3 had little or no pain at 7-9 years, although overall patient satisfaction was lower.

Cheilectomy does not restore normal joint kinematics. Nowoczenski (2008) found that the range of dorsiflexion was increased, but only to 31deg during gait, and the hallux remained planatar flexed during static stance. Heller and Brage (1997) reported abnormal sliding motions and early joint compression. Mulier (1999) reported that the area of maximum pressure was located under the first metatarsal head in 31% and post-operatively in 55%; however, Nawoczenski reprted that medialisation of the region of maximum pressure occurred only in 4/15 patients.

Cheilectomy has a definite role in treatment of mild-moderate hallux rigidus and is probably the procedure of choice in this group. Comparative trials with osteotomies of the phalanx and metatarsal would be useful in evaluating definitive best therapy. Cheilectomy is an acceptable option for Grade 3 hallux rigidus where the patient wishes to preserve movement and accepts the slightly higher failure rate, but not where there is pain throughout the range of movement.

Dorsiflexion phalangeal osteotomy

In patients with a reasonable range of residual motion, a dorsal wedge osteotomy of the phalanx increases dorsiflexion at a theoretical cost of loss of plantarflexion. Thomas and Smith also found that the dorsal space in the joint was increased by osteotomy. Originally this was described as an extra-articular procedure and this was said to be one of the advantages, but more recent series have combined the procedure with limited cheilectomies.

Four series, all with small numbers but two with follow-up of over 10 years, have reported satisfactory pain relief in 72-100%. Dorsiflexion increases by an average of 10deg, and Thomas and Smith (1999) also found a small increase in plantarflexion.

Phalangeal osteotomy has a role in management of mild-moderate hallux valgus. Comparative trials against cheilectomy are required.

Metatarsal osteotomy

The concept of metatarsus primus elevatus as a major aetiological factor in the development of hallux rigidus has led to the use of procedures which plantarflex the first metatarsal head, often with shortening to decompress the joint. Osteotomies have been performed proximally and distally, including the Youngswick procedure, which is basically a modification of the chevron osteotomy with a slice removed from the dorsal limb to slide the head down and proximally, and straight or oblique osteotomies of the head similar to the Weil procedure. The Waterman-Green osteotomy removes a wedge from the dorsum of the metatarsal head, aiming to rotate the better cartilage on the lower part of the head up into the metatarsal-phalangeal articulation.

These procedures are intended for use in early hallux rigidus and hence should be compared with the much simpler cheilectomy. Unfortunately, few results have been published (Roukis et al 2003, Gonzalez et al 2004) in these procedures, mostly with short follow-up. Reported results seem similar to those of cheilectomy, which is a simpler operation and, indeed, often forms a preliminary part of osteotomies.

Malerba (2008) reported 11-year follow-up of an oblique osteotomy of the metatarsal neck, which slides the head down and proximally (3mm on average). Criteria for surgery included grade 3 (advanced) hallux rigidus and metatarsal head elevation. In 23 feet in 20 patients, the mean AOFAS hallux score improved from 44 to 82/100 and the 1st MTPJ range of movement from 8 to 44deg. Joint space was maintained in 16 feet. Pedobarography showed maximum force on the distal hallux in 18 feet and on the lateral metatarsal heads in 5; one patient had transfer metatarsalgia.

The place of metatarsal osteotomies in the management of hallux rigidus is uncertain, but as they are more complex than the cheilectomy and based on a theory of the cause of hallux rigidus for which the evidence is at best equivocal, only good comparative studies will show if they have a useful place. Malerba's series at least gives long-term functional outcomes.

1st MTPJ arthrodesis

Approach is normally dorsomedial or medial, protecting the cutaneous nerve.

The joint is debrided of osteophytes and synovitis. Cannulated ball-and socket reamers or simple use of burr, rongeurs and drill produce a surface that can be positioned as required. Alternatively, a power saw can be used to produce flat cut surfaces in the desired position. The former technique is often said to produce less shortening of the ray, but Singh (2008) found no significant difference in first ray length in a cadaver model. Politi (2003) found that flat cuts produced a more stable construct than reamers with lag screw fixation.

Most authors recommend fusion in a position  of neutral rotation and about 10deg valgus. Dorsiflexion can be described relative to the sagittal plane position of the fitst metatarsal, but this does not allow for metatarsal declination. A simpler method is to position the toe in dorsiflexion such that, with the foot plantigrade, the pulp of the toe is just off the surface supporting the foot. A flat surface is useful to achieve this positioning.

Many different fixation techniques have been described, including wires, staples, one or two screws and low-profile plates. Politi (2003) found, in a cadaver study, that the most stable construct was a lag screw and dorsal plate. Faraj (2007) found screw fixation to be six times more stable than a circumferential wire in a nylon bone model. However, Hyer (2008) found that a plate fixation construct cost $603 as compared with $374 for crossed screws, with no difference in union rate, time or complications. Sharma (2008) found no difference in union or complication rates between a single lag screw and a lag screw and plate construct in 34 fusions, although this was probably underpowered to detect a plausible difference.

Aftercare described in published series has varied widely. Most series have maintained limited weightbearing on the heel or lateral foot border, or completely non-weightbearing, for 4-6 weeks. Mah (2008) allowed full weightbearing immediately in a padded post-operative shoe. Their union rate at 86% was lower than most modern series, but they used only K-wires for fixation. Given that weightbearing in a post-operative shoe does not produce excess failure in other forefoot procedures such as metatarsal osteotomy or Lapidus fusion, it seems unlikely that it would be a problem for 1st MTP fusion, but a RCT would be useful.

Many series of first MTP fusion include multiple pathologies such as hallux rigidus, hallux valgus and revision cases. Earlier series reported non-union rates of up to 40%, but recent series using compression screws and/or plates report 5% or fewer non-unions. Modern series report pain is improved in the majority of patients, with mean AOFAS forefoot scores 75-89/90 and satisfaction in 90% of patients. DeFrino et al (2002) found restoration of the weightbearing function of the first ray on gait analysis 3 years after fusion, with reduction in step length and reduction in ankle torque and power in late stance phase. Interphalangeal arthritis was not a problem at a mean of 6 years’ follow-up in Coughlin’s (2003) series.

Roukis (2011) summarised these and other studies in a systematic review that included 2656 procedures. The overall incidence of non-union was 5.4%, of which a third were symptomatic. Malunion occurred in 6.1%. Roukis commented on the need for more high-quality studies comparing techniques, and for high-quality studies concentrating on single indications for surgery.

A randomised controlled trial by O’Doherty et al (1990) found no difference in functional outcome between arthrodesis and Keller arthroplasty in 81 patients over the age of 45 (some of whom had hallux valgus in addition to OA). The technique of fusion in this series was not one that most surgeons would use but non-union made little difference to the result. The results of Keller's procedures were better in this study than in many others but this may reflect the standards which can be achieved with care.

1st MTP fusion is probably still the gold standard of care for advanced hallux rigidus. Alternatives include MTP replacement, capsular arthroplasty, Keller arthroplasty, metatarsal osteotomy and cheilectomy. However, there is insufficient comparative data on metatarsal osteotomy or capsular arthroplasty, and so far joint replacement has not been shown to give better functional results in RCTs.

Capsular arthroplasty

Hamilton et al (1997) described a capsular interposition arthroplasty for advanced hallux rigidus which included a limited resection of the proximal phalanx and suture of a dorsal capsular flap to the plantar structures - it may therefore be viewed as a modification of the Keller procedure. 34 feet in 30 patients were reviewed; duration of follow-up is not stated. There were improvements in the pain and function sections of the AOFAS hallux score and 28/30 patients were subjectively satisfied. Battaglia and Basile reported a technique which is not entirely clear but included limited resection and reshaping of the metatarsal head. At 5-year follow-up 91% had useful improvement in pain and function.

Capsular arthroplasty offers an alternative to arthrodesis, implant arthroplasty or Keller procedure for advanced hallux rigidus. Evaluation of its place in practice requires prospective randomised trials against one or more of these procedures.

Silastic arthroplasty

Silastic interposition arthroplasties were originally introduced to improve the stability of the Keller arthroplasty. Both single-stemmed and double-stemmed implants have been used. The Swanson double stemmed implants now come with metal grommets to reduce fretting against bone edges.

Silastic arthroplasty has a reputation for high wear, loosening, silastic synovitis and lymphadenitis and failure. Medium term studies in double-stemmed implants, however, have not borne this out, with fewer than 10% of implants revised at 5-9 years and little prosthetic subsidence or bone destruction (Sebold 1996, Bankes 1999). Single-stem implants, however, have a higher failure rate, with over 10% revised at 10 years, overall clinical failure in 25-35% of patients and granulomas in 70% (Broughton 1989, Rahman and Fagg 1993, Shankar 1995)

Double-stem silastic arthroplasty is an acceptable treatment for severe hallux rigidus. It is usually recommended for older, less active patients. Grommets may reduce the incidence of silastic synovitis but require better-designed studies to evaluate this further. Single-stem arthroplasty should probably be abandoned.

Other joint replacements

Metallic hemiarthroplasty has been around for over 40 years, but other joint replacements have recently been reported. Numerous implant types have been described:

Most series have used a dorsomedial approach, though a few have used the medial approach to avoid damaging the dorsomedial cutaneous nerve. Careful blunt dissection with scissors enables the nerve branches to be identified and avoided. A longitudinal incision is made in the capsule and the joint surface exposed.

Osteophytes are cleared from both joint surfaces. Konkel recommended a standard cheilectomy with hemiarthroplasties to avoid impingement and maximise range of movement. Adequate release of plantar adhesions also helps gain maximum range of movement.

Preparation and implantation depend on the implant and should be learnt from the manufacturer’s instructions and by attending appropriate training. Joint surfaces are usually excised with power tools using pre-set measurements. Medullary fixation of implants is achieved after preparation with reamers. It is important to ensure the correct joint tension with trial implants to get the right balance between stability and adequate range of movement. Additional plantar release or (in hemiarthroplasty) cheilectomy may be necessary to get the best range of movement. Most implants are uncemented and a small number are threaded.

MTPJ replacements are, in general, not designed to correct significant hallux valgus, and poorer results have been noted in a few series where metatarsus primus varus remained uncorrected. The surgeon may have to judge whether to correct metatarsus primus varus with a basal osteotomy.

Capsular and skin closure is in the surgeon’s normal fashion.

In the majority of series, bandages are applied rather than a cast. Some series avoided weightbearing on the forefoot for 2-3 weeks but most allow weightbearing with protection and early active and passive mobilisation of the hallux.

MTPJ replacement is aimed at obtaining pain control equal to that of a fusion, while preserving movement to allow better function + activities than a fusion. There is also some evidence that gait is abnormal after even a successful fusion, and it is hoped that MTPJ replacement will protect other joints from abnomal function and degeneration.

Several series of MTPJ replacements have been published using different designs. Cook (2009) carried out a systematic review and meta-analysis. However, they analysed subjective patient satisfaction as their only endpoint. While this is obviously important, they did not analyse pain, function, activities or work, and did not report complication rates. Moreover, while they included some relatively poor quality case series, they did not include either Gibson’s (2005) RCT against fusion or Raikin’s (2007) historic comparison with fusion. Interestingly, over half the papers analysed by Cook were on silastic arthroplasty. They did find that patient satisfaction was lower with silastic than with more modern designs, and that there was no difference in satisfaction according to age or underlying disease. Overall, 85.7% of patients were satisfied with their outcome, and this was increased to 94.5 in prospective series with >5y follow-up and little loss to follow-up.

Several series have indeed reported satisfactory outcomes, including Hassan (2006) and Arbuthnot (2008) with the MOJE ceramic prosthesis, Gupta (2008) and Danilidis (2010) with the ToeFit Plus total joint replacement, Taranow (2005) and Kissel (2008) with the BioPro hemiarthroplasty, Pulavarti (2005) with the BioAction total joint replacement, Sorbie (2008) with the Trihedron hemiarthroplasty and Konkel (2009 and several other reports on the same series) on the Futura hemiarthroplasty. Some of these series noted significant radiological lucencies around prostheses that did not affect satisfactory clinical outcomes, but were mostly quite short follow-up so could not exclude future clinical loosening. Other series reported significant failure rates, such as Barwick (2006) and Brewster (2010) with the MOJE and Sinha (2010) with the BioAction.

Two series have compared the results to fusion. Gibson (2005) reported an RCT between the Koenig total joint replacement and fusion with cerclage wiring. The main outcome measure was improvement in the visual analogue pain score and this was significantly better with fusion. Six prostheses were revised for loosening and eight others were loose. There were 6 delayed unions and 7 wound infections in the fusions. Patient satisfaction was also higher in the fusion group. Raikin (2007) compared 21 BioPro hemiarthroplasties with a comparable historical control series of 27 fusions. The main outcome measure was the AOFAS score, which improved from a mean of 36/100 in each group to 83 in the fusions and 72 in the replacements; the difference was highly significant. Five hemiarthroplasties were revised for fusion and there was one non-union in the fusions. A RCT comparing the MOJE with fusion was also run in the UK and this was stopped because of the rate of implant loosening and subsidence; the clinical results have not yet been published but were not dissimilar to Gibson’s.

In view of the large variability in results and unfavourable results of comparative studies, we feel that 1st MTPJ replacement should currently only be performed in the context of a well-designed clinical trial.

Keller procedure

The Keller procedure is an excision arthroplasty of the 1st MTP joint, which can be used for both hallux rigidus and valgus. Most series contain a mixture. It has the potential to create an unstable toe, particularly if bony resection is excessive or soft tissue repair neglected, but can give useful results in carefully selected patients if a good soft tissue repair is carried out.

Approach is normally dorsomedial or medial, protecting the cutaneous nerve. The capsule is opened longitudinally and flaps are carefully elevated to expose the joint and the proximal 1/3 of the phalanx.

The joint is debrided of osteophytes and synovitis. The proximal 1/3 of the proximal phalanx is excised, usually with a power saw. A careful repair of the capsular/periosteal sleeve is carried out, taking care to repair the transverse tie-bar of the forefoot at the correct tension. A release of the extensor hallucis brevis is sometimes required to prevent “cock-up” deformity. Skin closure is as standard. The patient can mobilise full-weightbearing immediately. There are no trials of different methods of aftercare. Strapping of the toe is common, aiming to achieve healing with a stable toe.

The Keller procedure was for many years the main procedure for hallux valgus in many surgeons' armamentarium, especially in the UK. It attained a reputation for defunctioned great toes, "cock-up" deformity and metatarsalgia. It has been argued that this reputation is at least partly due to a lack of attention to correct technique, especially soft-tissue balancing, and to many procedures being done by inadequately trained and supervised junior surgeons.

Most series (eg O'Doherty et al 1990, Blewitt and Greiss 1993) are mixed populations of hallux rigidus and valgus. With this reservation, O'Doherty (1990) reported an RCT comparing Keller arthroplasty with fusion for arthritic 1st MTP joints, mostly with valgus deformity. Both procedures were equally effective in relieving pain in the 1st MTP joint and shoe fitting problems. Schneider and Knahr (2002) compared the results obtained by two surgeons in the same department in hallux valgus with HVA < 40deg, IMA <22deg and grade 1-2/3 degenerative change. One team used distal chevron osteotomy, the other Keller, and both carried out "cerclage fibreux" - lateral soft tissue release and medial capsular plication. The groups were similar except that the Kellers were slightly older, and follow-up was 5.7y. The mean final AOFAS hallux score was 85 in the Keller group and 83 for chevron. Radiological correction was comparable and there was only one cock-up toe in the Keller group. Metatarsalgia occured in 22% of chevrons and 20% of Kellers, and 92% of Keller patients and 94% of chevrons were satisfied. Beertema (2006), in a retrospective study of patients at all grades of hallux rigidus, found better results in grade 3 from Keller than fusion. A further RCT comparing Keller's procedure with fusion in older patients would be valuable.

The Keller procedure remains an useful treatment for end-stage hallux rigidus, especially in older, low-demand patients, diabetics (who tend to get stiff joints anyway) and as a salvage procedure. Further trials against better methods of arthrodesis and against silastic and other implants are required.