The scarf is a metaphyseal osteotomy of the first metatarsal. It is a very flexible osteotomy and can be modified to deal with many common problems. The flat osteotomy plane means that immediate post-operative weightbearing is possible without a cast. Some surgeons use it as their only, or main procedure for hallux valgus.
Access to the 1st MTP joint and to the osteotomy site is obtained from a medial longitudinal incision on the first ray from the middle of the distal phalanx to the first TMT joint. A lateral release of the first MTPJ may be done across the 1st MTP joint or through a separate longitudinal incision in the first intermetatarsal space.
We normally do the lateral release first through a 1st space incision. Dissection is deepened bluntly with scissors until the adductor hallucis tendon and the lateral capsule of the 1st MTPJ are exposed. The deep peroneal nerve and its branches are protected with retractors. A partial lateral capsulotomy is performed and the sesamoid-metatarsal ligament released. A Macdonald elevator in the 1st MTPJ is useful to identify and tense the sesamoid-metatarsal ligament for incision. The adductor tendon does not always have to be released (Schneider 2007), but if the contractures is very tight it can be released from the proximal phalanx taking care not to damage the flexor hallucis brevis.
A medial longitudinal incision is deepened to the capsule of the 1st MTPJ and the metatarsal shaft, obtaining any necessary haemostasis. The plantar part of the capsule is partially exposed to facilitate later double-breasting. The capsule is opened longitudinally and the medial eminence exposed. The eminence is excised with a power saw in line with the medial surface of the metatarsal shaft (not according to the groove in the metatarsal head, which will usually result in excessive excision). The joint is inspected and its alignment relative to the metatarsal shaft (DMAA) assessed.
The longitudinal cut is made first. It starts about 1cm proximal to the TMTJ, where the metatarsal starts to flare out, about 2mm above the junction of the superior and inferior surfaces (this is easier to identify on some metatarsals than others!) It runs distally parallel with the ground to 2-3mm short of the articular surface. The osteotomy plane runs downward parallel to the inferior surface of the metatarsal. It may be best to cut the medial surface completely first to establish the line.
Transverse cuts are then made from the ends of the longitudinal cut to the lateral surface, taking care to cut down in the same plane as the main cut. There is varying advice about the alignment of these cuts in the transverse plane. One suggestion is to make the cut at right angles to the long axis of the foot, which allows sliding without overall change in length. The more it is intended to rotate the distal fragment to correct the DMAA, the more proximally the transverse cuts need to be directed. Lengthening can be achieved by directing the cuts more distally, but is usually easier by physically translating the distal fragment distally.
The distal fragment is carefully separated and translated laterally by a combination of pushing the head and pulling the shaft fragments. DMAA can be corrected by rotating the distal fragment. Alternatively, the distal fragment can be rotated laterally, giving an effect like a proximal or Ludloff osteotomy (Larholt 2010). The metatarsal can be shortened if necessary to allow full correction of a stiff MTPJ, or to take some tension out of a degenerate joint, by removing a slice from each transverse cut. The downwards translation of the head allows more shortening than is the case with some other osteotomies. Alternatively, length can be gained by translating the distal fragment distally. Bone graft may be inserted in the gaps (Singh 2009)but is probably not necessary.
The osteotomy is fixed with two cannulated headless screws. Mini-fragment cortical screws can be used and are cheaper but patients sometimes feel the heads uncomfortable. The medial spike is trimmed off and any necessary rounding done. The capsule is repaired to restore soft tissue tension. We use a double-breasting technique. Once the correction is completed, or earlier if more convenient, the degree of residual hallux valgus in the phalanx can be assessed and, if necessary, and Akin osteotomy can be added. Some surgeons use an Akin osteotomy on almost every patient while others are selective; there is no evidence that routine Akin is necessary to obtain good results (Malviya 2007). Skin closure is as standard for the surgeon.
The osteotomy is stable to vertical loading and patients can weightbear immediately. Some series have used strapping or splintage post-operatively to minimise the risk of recurrent deformity. There are no trials of aftercare methods. Patients may be able to start using soft shoes with plenty of room within 2-3 weeks of surgery, although there tends to be more swelling than with a distal osteotomy.
The osteotomy usually takes 6-8 weeks to unite. Most surgeons take check Xrays early and then around the expected date of union to document initial correction and position of union. However, Murphy (2007) found that only one of 412 post-operative Xrays was noted to be abnormal, and no change in management was made.
Holt (2008) found that braking reaction time had returned to normal 6 weeks after a chevron, scarf or basal first metatarsal osteotomy. Although they did not measure braking response between 2 weeks (when most patients were unable to do the test due to pain) and 6 weeks, it is probably advisable to recommend that patient wait till 6 weeks before driving until further evidence is available.
The scarf is a flexible osteotomy which has become the “workhorse” of many surgeons’ practice. Nevertheless, it is quite a complex osteotomy and has unique problems (Coetzee 2003, Smith 2003), especially stress fracture related to the proximal end of the cut, and "troughing". It can be difficult to salvage if it goes wrong. Murawski (2011) recommended rotating rather than displacing the distal fragment to prevent troughing, but there are no comparative studies.
In a RCT, Deenik (2007, 2008) compared the chevron osteotomy to the scarf in 136 cases. There were no differences in radiological or clinical outcomes, even in severe deformities. The study design was altered to give more information about the results in severe deformities, which could have introduced some bias.
Smith (2012) performed a systematic review of studies of the scarf and chevron osteotomies. They found tha thte scarf corrected IMA by 0.88deg more than the chevron. They suggested that the methodological quality of the evidence was such that this apparent difference was not very reliable.
Robinson (2009) reported a sequential comparative study of the scarf and Ludloff osteotomies. The scarf group were done first so had longer follow-up. The scarf group had more pain and more problems with swelling and shoe wear pre-operatively. Intra-operative correction was equivalent in both groups but better maintained after a scarf. 68% of pre-operative transfer lesions resolved after a scarf compared with 58% after a Ludloff, and there were no new transfer lesions. There were no non-unions or mal-unions in the scarf group compared with three in the Ludloff group. However, there were no significant differences in a number of functional assessments between the two groups.
Dhukaram (2006) compared both clinical and pedobarographic results of the Mitchell and scarf osteotomies retrospectively. The scarf group had higher AOFAS forefoot scores, fewer lesser ray calluses, less transfer loading of the 2nd and 3rd rays and less MTPJ pain.
Weil, who popularised the scarf osteotomy in the 1990s, published a large, long-term series (Weil 2000), but without a great deal of detail and no comparison between the standard procedure and the modifications which he recommended.
Kristen (2002) reported 111 scarf osteotomies in 89 patients followed up for 2-4y. The mean AOFAS hallux score was 91/100. 90% of patients had little or no pain. Patients returned to work in a mean of 5.8 weeks and to sports in a mean of 8.4 weeks. Seven patients had recurrent hallux valgus and two had wound infections.
There have been a number of smaller studies (Perugia 2003, Aminian 2006, Berg 2007) with follow-up under 5 years, mostly reporting satisfaction rates around 85-90% and complication rates 5-10%.
George (2009) found a high rate of recurrence of hallux valgus deformity after 3 years in adolescent hallux valgus.
The scarf comes out fairly well from the reported series. However, more comparative studies, especially well-designed RCTs, would clarify its place further.