Clinical outcome of nerve injuries associated with supracondylar fractures of the humerus in children

Neurological complications associated with supracondylar fractures of the humerus in children are well recognised.^1,2 The relative incidences of traumatic and iatrogenic nerve injuries associated with this fracture have been reported as being 12% to 20%^1,3,4 and 2% to 6%,^5,6 respectively. The radial and anterior interosseous nerves are thought to be those most commonly involved by the fracture itself^2,3,7 while iatrogenic damage most often affects the ulnar nerve.^8–10

Several studies have suggested that 86% to 100% of these nerve injuries are neurapraxias which resolve spontaneously within six months, with the mean time to recovery being between two and three months.^3,4,8,11 By contrast, other reports have noted inadequate recovery and the need for surgical intervention in selected cases.^1,6,9

Our aim was to report the referral pattern and the outcome of nerve injuries associated with supracondylar fractures of the humerus in children seen over a period of five years at a dedicated peripheral nerve injury unit in the UK.

Patients and Methods

The database of our peripheral nerve injury unit was reviewed to identify all patients under the age of 16 years who had been referred for the further management of a nerve injury associated with a supracondylar fracture of the humerus between January 1, 1998 and December 31, 2002.

Children with other injuries to the elbow, such as condylar fractures and physeal injuries, were excluded from the study.

We identified 32 patients with 32 fractures and 37 nerve injuries. There were 19 boys and 13 girls with a mean age at referral of 7.9 years (3.6 to 12.5; Table I). The mechanism of injury in 31 patients was a fall from a height. The remaining patient (case 18) had been the front-seat passenger in a car involved in a road-traffic accident. All the fractures were closed, displaced extension-type injuries. The fractures were classified using the Gartland classification at the referring hospital¹² (Table II), and this was confirmed, when possible at the first consultation in our unit. Eight of the 32 fractures were classified as Gartland type II and 24 as Gartland type III. At the referring hospital, two of the fractures had been treated by closed reduction and above-elbow casting, 20 by closed reduction and percutaneous cross Kirschner (K-) wire fixation with a mini-open medial approach to protect the ulnar nerve, and ten by formal open reduction and cross K-wire fixation. A lateral approach to the fracture had been used in nine of ten open reductions, with an anterior approach in the final case to allow concomitant exploration of a vascular injury. One further patient had an anterior incision for a delayed exploration for vascular compromise (Table I). The neural injuries were initially managed by observation at the treating hospital, and referred when inadequate recovery was noted.

All children presented to our unit within three months of the injury. The indications for operative intervention were either a complete degenerative lesion on neurophysiological studies or failure of the anticipated clinical recovery. The operative findings and details of the surgical procedure were recorded. Post-operative follow-up was carried out at six weeks, three months and subsequently every three months until full recovery. The clinical outcome at the last follow-up was assessed according to the criteria of Birch, Bonney and Wynn Parry¹³ and graded as excellent, good, fair or poor (Table III).

Results

Clinical details are shown in Table I. Nineteen (51.4%) of the referred nerve injuries involved the ulnar, ten (27%) the median and eight (21.6%) the radial nerve. Out of 37 neuropathies 14 (37.8%) were directly related to the injury itself and were noted at the time of presentation. The remaining 23 (62.2%) were diagnosed after reduction and fixation. Of the 14 neuropathies related to the injury, five were to the ulnar, five to the median and four to the radial nerve. Of the 23 diagnosed after initial intervention, 14 were to the ulnar, five to the median and four to the radial nerve. Nine of the 14 (64%) injuries to the ulnar nerve were associated with closed reduction and percutaneous pin fixation of the fracture. There were no cases of compartment syndrome. In two of the 32 patients, there had been a suggestion of vascular compromise. One case was noted at presentation and an anterior approach was used with crossed K-wire fixation of the fracture (case 16) and in the other, vascular compromise was noted after closed reduction and percutaneous fixation and an anterior approach to the vessels was performed 24 hours later (case 29).

Spontaneous neurological recovery occurred in 19 patients (24 nerves) at a mean of 7.7 months (3 to 15) after injury, with 16 patients having an excellent and three a good outcome.

Operative intervention was planned for 13 patients (13 nerves) either because of a complete degenerative lesion or because of clinical failure of recovery (Table IV). In three, surgery was cancelled on the day of admission as there were signs of recovery of nerve function and two of these recovered fully by six months from the time of the injury, while the third recovered by nine months. All had an excellent outcome.

Ten patients underwent exploration of their nerve lesions at a mean of 7.7 months (6 to 9) after injury. At exploration, nine nerves were noted to be in continuity. A nerve stimulator was used to assess functional continuity across the site of the injury. When distal function was noted with proximal stimulation, conservative surgery was performed.

An external neurolysis was performed in six ulnar nerves, four were trapped within callus and two in fibrous scar tissue within the cubital tunnel. Four patients had an excellent and two had a good outcome.

The remaining four nerve lesions (two radial, one ulnar and one median) required nerve grafting to establish functional continuity, three because of the formation of a neuroma with no distal response to proximal stimulation, and one because of laceration, retraction of nerve ends and resultant gap. The medial cutaneous nerve of the forearm was used as the donor in three patients, and the superficial radial nerve in one. There were no operative complications. Of these four patients, three had an excellent, and one a fair outcome. The last child did not regain extension of the thumb after a radial nerve injury treated by superficial radial nerve grafting. She subsequently underwent transfer of palmaris longus to extensor pollicis longus.

Although 50% (four of eight) of the lesions of the radial nerve occurred in association with a lateral approach for an open reduction only two of the four were iatrogenic. All four recovered spontaneously although surgery was planned in two.

Overall, 26 patients (81.3%) had an excellent, five (15.6%) a good and one (3.1%) a fair outcome. In 13 patients 14 lesions were noted at presentation of the fracture; two of these patients subsequently underwent surgical exploration. Of the 13 patients, 12 (92%) had an excellent outcome. Twenty-three nerve lesions in 20 patients were classified as iatropathic in that the nerve injury was noted after intervention either in terms of a manipulation or percutaneous fixation or open reduction. One patient (case 19) had a lesion of the median nerve identified at presentation and injury to the ulnar and radial nerves after open reduction and percutaneous fixation. Eight patients (40%) with iatropathic injuries underwent a surgical procedure. Of the 20 patients 15 (75%) had an excellent outcome.

Discussion

Nerve injuries after supracondylar humeral fractures occur primarily due to tenting or entrapment of the nerve on the sharp proximal humeral fragment, while iatrogenic injuries occur either during closed manipulation or percutaneous fixation of the fracture fragments or occasionally during open procedures.^8,9,14 The median or anterior interosseous nerves are most commonly damaged by extension-type injuries, while the less common flexion-type injuries affect the ulnar nerve more often.^15–17

It has been reported that iatrogenic injuries most commonly affect the ulnar nerve, with percutaneous crossed K-wire fixation being associated with the highest risk of injury.¹⁰ In a smaller study by Green et al,¹⁸ only one case of neurapraxia of the ulnar nerve was identified in 65 patients treated in this way. This study which was based on late referrals to a specialist nerve injury unit similarly showed a relationship between lesions of the ulnar nerve and closed reduction and percutaneous crossed K-wire fixation. Our unit accepts referrals from all over the country. We cannot comment on the true incidence of supracondylar-associated neural injuries since referrals were only made to us by the treating hospitals when the initial period of observation had failed to show recovery. Therefore, transient neurapraxias, whether traumatic or iatrogenic in origin, are not included in our figures. This may explain the low incidence of anterior interosseous nerve lesions in our study. In addition, we assume that some neural injuries would have been treated locally by other appropriate units.

Most studies in the orthopaedic literature have reported a good to excellent prognosis for nerve injuries associated with supracondylar fractures in children.^3,4,8,11 A few have emphasised the need for surgical intervention in selected cases.^1,6,9 Culp et al¹ reported 18 nerve injuries in a retrospective review of 101 displaced extension-type supracondylar fractures. They explored nine cases when there was no clinical or electromyographic evidence of return of function at a mean 7.5 months after injury; eight lesions were in continuity, with nerve function recovering well after neurolysis. One complete laceration of the radial nerve, however, did not recover even after grafting and underwent tendon transfers.

Birch and Achan⁶ reported 118 cases of repaired nerve lesions “associated with fractures and dislocations at the elbow displayed at operation”, of which 91 were in conjunction with supracondylar fractures. Of these, 43 involved the median, 35 the ulnar and 13 the radial nerve. Seven cases were iatropathic, although it was unclear if any or all of these were related to the supracondylar fractures. They specified that 22 of the 91 injured nerves associated with supracondylar fractures were found to be entrapped within the fracture or impaled on a bone spike, while the remainder were compressed by swelling or fibrosis at or distal to the fracture, the latter showing uniformly good recovery after decompression. No reference was made to the need for grafting.

In regard to iatrogenic injuries, the reported prevalence of injury to the ulnar nerve with the use of crossed K-wires has ranged from 2.5% to 6%.^8–10 Brown and Zinar,⁸ in reviewing 162 fractures, identified four iatrogenic lesions of the ulnar nerve. All were explored and in all the medial wire was found to be injuring the nerve. They recommended urgent re-exploration in cases of post-operative nerve palsy. By contrast, Lyons et al⁹ reported injuries to the ulnar nerve noted post-operatively after percutaneous cross K-wire fixation in 375 Gartland type-III supracondylar fractures. Of these cases 17 were followed up and all did well irrespective of whether the wire was removed, the nerve explored or the patient treated conservatively. The authors comment that placement of the wire is only one of several factors implicated in the development of a post-operative nerve palsy. It is common practice now to ensure that the medial wire is placed through a mini-medial approach in an attempt to reduce the risk of nerve injury. Skaggs et al¹⁹ reported no significant reduction in iatrogenic injuries to the ulnar nerve when such an approach was used. By contrast, Green et al¹⁸ reported only one case of transient sensory symptoms in the ulnar nerve in a two-surgeon series of 65 patients treated using a mini-incision technique. In all the cases in our series the medial wire had been introduced through a mini-medial incision. Rasool⁵ documented nerve injury occurring secondary to constriction by the cubital tunnel retinaculum. Cases 12 and 26 were examples of this phenomenon in which scar tissue in the cubital tunnel had caused a complete degenerative lesion of the ulnar nerve, with a good to excellent outcome after neurolysis. Skaggs et al¹⁹ reported no iatrogenic injury to the ulnar nerve when only lateral wires were used, and confirmed that adequate fixation of unstable supracondylar fractures was achieved by such a technique. De las Heras et al²⁰ identified two lesions of the ulnar nerve in 77 patients using a lateral technique and emphasised the need for stability of the fracture, if necessary using a third wire, to reduce the risk of nerve injury. None of these studies reported the incidence of traumatic nerve injury documented at presentation. As pointed out by Lyons et al⁹ the observation of nerve injury post-operatively does not imply that it occurred as a result of the intervention – it is possible that it had simply not been recognised at the time of presentation. Similarly, the trauma associated with the reduction may injure the nerve irrespective of whether percutaneous wiring techniques are used. In our patients all the wires had been removed before referral to our unit and we are unable to comment on whether or not the nerve lesions were due to penetration by the wires or damage associated with their introduction. Only two wires in two patients had been removed ‘early’ because of the presence of a nerve palsy. Cases of direct damage by K-wires may have been recognised and treated locally and if recovery of nerve function followed they would not have been referred to our unit. Of the ten patients who underwent exploration, one neuroma appeared to be related to the site of insertion of the wire but most nerve lesions were secondary to entrapment within fracture callus or scar tissue.

A conduction block may progress to a degenerative lesion if the compressive element of the original injury is not relieved. The presence of persistent neuropathic pain implies ongoing nerve compression and injury and should dictate further investigation and relief of compression. Clinical examination should also concentrate on sympathetic function. A warm dry digit implies that there is nerve dysfunction and indicates the need for exploration of the nerve. Our suggested indications for exploration are given in Table V.

We conclude that contrary to the implications in the recent literature that nerve injuries associated with supracondylar fractures invariably recover spontaneously and well, certain nerve injuries with complete lesions on neuro-physiological studies and those in whom the anticipated clinical recovery does not occur require assessment and management in a specialist unit.

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.