Nerve injuries associated with supracondylar fractures of the humerus in children

Supracondylar humeral fractures are the most common elbow fractures in children, representing 17.9% of all paediatric fractures.¹ They have a male preponderance, with a male:female ratio of 6:4.^2,3 The peak incidence is between five and seven years of age.² Supracondylar fractures can be described as either extension or flexion injuries, according to the fracture pattern and the mechanism of injury. The majority are extension-type injuries, comprising 98% to 99%,^3-6 these are typically sustained by falling onto an outstretched hand with the elbow in full extension.³ The Wilkins-modified Gartland classification⁷ further divides this group into three sub-types, according to the degree of displacement⁸ (Table I).

Supracondylar fractures can be significant injuries with a reported incidence of associated nerve injuries between 11% and 15%.^4-6,9 Posteromedially displaced fractures are more likely to have an associated neurological injury;¹⁰ most commonly the radial nerve is affected. Posterolateral displacement of the fracture fragment is more likely to injure the median nerve.^10-12

Nerve injury was originally classified by Seddon in 1943¹³ and subsequently modified by Sunderland in 1951.¹⁴ The least severe injuries are neurapraxias (Sunderland grade I), these are caused by transient ischaemia and lead to a loss of superficial sensibility, followed by pain and loss of motor function. Segmental demyelination occurs, leading to a local conduction block. Large myelinated fibres are affected first, followed by C-fibres. In axonotmesis (Sunderland grade II), the axon is disrupted but the Schwann tubes remain intact. Complete Wallerian degeneration takes place in the subsequent days; the axon degenerates distal to the site of injury and the granular disintegration of the cytoskeleton and axoplasm becomes amorphous debris.¹⁵ The integrity of the Schwann cell basal laminae are crucial for axonal regeneration. Neurotmesis (Sunderland grade III-V) occurs when axonal interruption is associated with transection of the basal laminae; as all elements of the nerve are in discontinuity, recovery is only possible if the nerve is surgically repaired.

In 1998, Birch, Bonney and Parry¹⁵ devised a simpler classification of nerve injuries into ‘degenerative’ and ‘non-degenerative’ lesions. As this can be categorised according to clinical findings, it is useful in clinical practice and has prognostic value. Non-degenerative lesions are neurapraxias caused by a conduction block, the effects of which may last for weeks or months following removal of the offending cause. No Wallerian degeneration occurs;¹⁵ recovery is usually complete and the prognosis is good in this group. Degenerative lesions encompass axonotmesis and neurotmesis. Proximal to the site of injury, the interrupted axon forms multiple axon sprouts with a growth cone at the tip of each sprout. Percussion along the nerve, from distal to proximal, stimulates the axon sprouts and manifests as Tinel’s sign, where a surge of ‘pins and needles’ or abnormal painful sensations are experienced in the dermatomal distribution of the injured nerve. The prognosis for a degenerative lesion is generally less favourable. The key difference between a non-degenerative and degenerative lesion is the persistence of conduction in the distal nerve segment in the former and loss in the latter,¹⁵ this can be demonstrated in nerve conduction studies.

In this study we consider the pattern of nerve injury, treatment and clinical outcomes in children who were referred to our unit following supracondylar fractures over a period of 17 years. This expands considerably on earlier work published from our unit.¹⁶ We discuss the clinical signs and symptoms that should prompt concern and define our indications for surgical exploration.

Patients and Methods

We retrospectively reviewed the records of all children with nerve injuries, associated with supracondylar humeral fractures, referred to our peripheral nerve injury unit between January 1996 and December 2012. The patients were identified through our database. Permission was granted by the unit for review of the database. Inclusion criteria were: a diagnosis of supracondylar fracture, the presence of neurological injury or dysfunction and age under 16 years. We excluded other injuries around the elbow, for example elbow dislocations, condylar fractures and olecranon fractures. We obtained clinical data from case notes, imaging studies and outpatient clinic reviews. We identified 166 patients, of whom 111 were male and 55 were female. The mean age at the time of injury was seven years old (standard deviation (sd) 2.2). We recorded the mechanism of injury, neurological and vascular status for each patient. Extension-type injuries were sub-classified according to Gartland’s system.⁸

In all cases, the initial fracture management was at the referring hospital. After review in our outpatient department, patients were managed either conservatively, or operatively with exploration and neurolysis of the affected nerve(s). Further procedures were carried out depending on the intra-operative findings (for example, nerve repair, grafting or transposition, removal of metalwork, revision of fracture fixation).

We classified the nerve injury into non-degenerative or degenerative lesions, according to Birch et al¹⁵ using a combination of clinical features, neurophysiology data (where available) and intra-operative findings. Degenerative lesions were further subdivided into partial and complete injuries. We described the nerve injury as iatrogenic when the documented neurological status deteriorated after their initial fracture management surgery (e.g. from entrapment of the nerve by reduction of the fracture), or when there was clear intra-operative evidence that the nerve was likely to have been damaged by Kirschner (K)-wires (e.g., penetration of the wire through the nerve or epineurally).

The mean follow-up time was 12.8 months (four days to 104 months). Motor power was assessed using the Medical Research Council (MRC) scale.¹⁷ The final clinical outcome was recorded by the Birch et al¹⁵ modification of the original Seddon system.

Results

The mean time from injury to referral to our specialist unit was 89 days (one day to three years seven months).

Injury patterns

Of the 166 patients reviewed, 165 (99.4%) were extension-type injuries and only one was a flexion-type injury. All but one patient sustained their supracondylar fracture through falling on an outstretched arm, with 22 (13.3%) children falling from a climbing frame and 14 (8.4%) from a trampoline. The demographic data of the 166 patients are shown in Table II.

A total of 196 nerves were injured in the 166 children; 26 had neurological dysfunction affecting two or more nerves. The most commonly injured nerve was the ulnar nerve (43.4%), followed by the median (36.7%) and radial nerves (19.9%). The distribution of affected nerves is illustrated in Figure 1.

Vascular injury

Of the 33 children who sustained a concurrent vascular injury, eight had required immediate vascular exploration and intervention at the referring hospital. Initially 25 patients had a ‘pink, pulseless hand’, two regained pulses after reduction by manipulation alone and the remainder required closed or open K-wire fixation. Only 19 of the 33 patients required surgical exploration for their neurological injury at our unit.

Operative findings

Following assessment in our unit, a total of 94 children (56.6%) required surgical exploration of their nerves. Their operative information is presented in Table III.

The remainder were treated non-operatively. Of the 72 children who did not require operative intervention, 16 were initially to have surgical exploration, however, this was cancelled on admission as they exhibited significant clinical recovery.

Of the 30 patients with iatrogenic nerve injuries, 18 were related to K-wire use, either due to nerve penetration, epineural stripping or tethering. The remainder (n = 12) were trapped in the fracture site at the time of reduction.

Clinical outcomes

Clinical outcomes at follow-up outpatient review of all 166 patients are shown in Table IV. Further surgery was required in eight children and ten suffered complications from their initial injury. Table V shows the treatment and clinical outcomes of patients according to whether they had a positive Tinel’s sign on initial presentation.

Discussion

Supracondylar fractures of the humerus are frequently associated with nerve injuries in children, occurring in between 11% to 15% of cases.^4-6,9 However there appears to be no clear consensus as to which nerves are most commonly affected, with different investigators reporting different results.^5,12,18-20 In this series of 166 children referred to a tertiary centre, the ulnar nerve was the most commonly injured. A total of 26 children (15.7%) were found to have neurological dysfunction in two or more nerves. The majority, 120 patients (72.3%), had a Gartland Type III fracture.

Concurrent vascular injuries were seen in 33 patients, nearly 20% of our cohort. In the ‘pink, pulseless hand’, a concurrent nerve palsy prompts early exploration as it is strongly predictive of nerve and vessel entrapment.²¹ This is in line with the recently published British Orthopaedic Association’s guidelines, which recommend urgent surgical intervention in the absence of a radial pulse, or with clinical signs of impaired perfusion to the hand and digits.²²

There were eight ‘anterior interosseous nerve’ (AIN) palsies with pure motor dysfunction in the radial part of the flexor digitorum profundus and flexor pollicis longus. This is somewhat of a misnomer, as the AIN arises away from the site of injury in supracondylar fractures, from the radial aspect of the median nerve 2 cm to 6 cm distal to the medial epicondyle of the humerus.⁴ The association of AIN palsy with supracondylar fractures was first described by Lipscomb and Burleson in 1955¹⁹ and later by Spinner and Schrieber in 1969.²³ The proposed mechanism by the former was direct trauma at the fracture site to the posterior moiety of the median nerve (proximal to the AIN).¹⁹ Cross-sectional mapping of the median nerve by Sunderland, however, demonstrated that the AIN fibres were situated immediately adjacent to the median sensory fibres.²⁴ Spinner and Schreiber²³ therefore pointed out that:

“blunt trauma to the posterior aspect of the median nerve sufficient to produce total paralysis of these muscles should not completely spare the sensory fibres in immediate proximity of these motor fibres.”

They proposed an alternative mechanism of injury, rather than pressure from posterior displacement of the fracture fragment, whereby traction damage is caused to the AIN fibres due to its relatively fixed position in the proximal forearm. Our experience, with intra-operative observation of these injuries, suggests focal damage rather than a traction model. We note that partial sensory disturbance is difficult to assess acutely in this patient group; this requires further study.

In terms of intra-operative findings, the most common cause for nerve dysfunction was entrapment or tethering of the affected nerve, either by the fracture itself, within the elbow joint, by fibrous scar tissue or K-wires. Nerve lacerations were rare, particularly complete transection (0.08%); this is in keeping with previous literature.^10,25

Neurological dysfunction following supracondylar fractures in children is generally thought to be neurapraxia that will resolve spontaneously in two to three months.^4,5 In this series, we have shown that only 54 patients (27.5%) referred to our tertiary specialist centre had non-degenerative injuries. In the diagnosis of a nerve lesion, clinical examination findings are important. The presence or absence of Tinel’s sign (which was documented in only 106 (63.9%) patients’ clinical notes in this study) is a valuable clinical indicator in determining whether a nerve injury is degenerative or non-degenerative in nature. Percussion should track the surface marking of the nerve from distal to proximal. The level of a positive Tinel’s sign should be measured as a distance from a fixed bony landmark and clearly documented in the patient’s clinical notes. In subsequent examinations, a regenerating, degenerative nerve lesion would be expected to have an advancing Tinel’s sign. This is useful in the prediction of neurological recovery and prognosis; degenerative lesions of favourable prognosis (axonotmesis) advance more rapidly, at up to 2 mm a day, compared with those of an unfavourable prognosis.¹⁵ In principle, a non-advancing Tinel’s sign is an indication for nerve exploration.²⁶ Our incomplete data set in respect of the Tinel’s sign is inconclusive though, and further work is required to confirm the predictive value in these injuries.

The mechanism of fracture reduction and fixation is out of the scope of this series, but pragmatically the authors do not think that it is realistic to perform open reduction and internal fixation in all of these injuries. This is a series of significant injuries which have been referred to a tertiary centre, and are more likely to represent injuries which were initially more severe.

The presence of neuropathic pain is often difficult to identify in children. It may manifest as an exaggerated aversion to light touch of the hand or forearm by the parent or clinician, even weeks or months following the initial injury. Sympathetic dysfunction can be observed by swelling, discolouration and anhidrosis in the hand. If the sympathetic supply to the hand is intact, sweating and wrinkling of the skin will be elicited by wrapping the child’s hand in a damp, warm towel. In the presence of sympathetic dysfunction, a degenerative lesion is most common.²⁷ In total 82 of the patients referred had neurophysiological investigations (49.4%). The predictive value of this is outside the scope of this review, but serial examination and clinical assessment remain the most important modalities that guide treatment.

In this series, 94% of children (154 patients) achieved an excellent or good final clinical outcome, highlighting the importance of prompt and appropriate treatment. Certainly if the patient shows little sign of clinical or neurophysiological recovery, they shuld be promptly referred to and explored in a specialist unit; in nerves with good anatomical continuity, neurolysis alone can lead to excellent recovery by relieving compression from fibrous scar tissue.^5,28,29 In operated patients, the outcomes were overall good, (Table IV) but there were more results in the poor outcome bracket compared to the non-operative group. This may reflect the greater number of patients with associated vascular injuries in the operative group. The detailed analysis of this group requires further study. Even in the 13 children (7.8%) who required nerve grafting in this series, ten had excellent or good outcomes after surgery.

We identified 30 iatrogenic injuries, over half of which were related to K-wires, the remainder to fracture reduction. Following our surgical intervention, all patients achieved excellent or good outcomes and only eight patients (4.8%) required further surgical procedures to manage the sequelae of their injury. One patient required a wedge osteotomy of the distal humerus in order to correct a cubitus valgus deformity, a recognised consequence of supracondylar fracture malunion.^30,31

In conclusion, all three major nerves (median, ulnar and radial) are associated with supracondylar fractures in children. In the context of complete palsy, the presence of a positive Tinel’s sign, neuropathic pain or vascular compromise, prompt referral to a specialist unit for consideration of nerve exploration is recommended. Where managed appropriately, nerve recovery and clinical outcomes are extremely favourable for this paediatric population.

Author contributions:

I. H. Y. Kwok: Study design, Data acquisition; Data interpretation; Data analysis; Writing of manuscript, Manuscript revisions.

Z. M. Silk: Data interpretation, Data acquisition, Review and editing of manuscript.

T. J. Quick: Performed surgeries, Review and editing of manuscript.

M. Sinisi: Performed surgeries, Review and editing of manuscript.

A. Macquillan: Performed surgeries, Review and editing of manuscript.

M. Fox: Performed surgeries, Study design, Data interpretation, Review and editing of manuscript, Manuscript revisions.

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.

This article was primary edited by E. Moulder and first proof edited by G. Scott.

Supplementary material. Tables showing two grading systems are available alongside the online version of this article at www.bjj.boneandjoint.org.uk.