12.35 Humeral shaft fractures

Humeral shaft fractures are commonly closed injuries which will generally unite with non-operative treatment with good clinical function. A small proportion have an associated injury to the radial nerve, but this lesion is usually in continuity, and the role of early exploration of a nerve lesion present from the time of a closed fracture is controversial. The treatment options and their rationale are discussed.

In a Swedish study, humeral shaft fractures were seen to occur in 14.5 per 100 000, the incidence increasing with age. While in this study the majority were due to simple falls, high-energy trauma, penetrating trauma, and the indirect trauma of throwing and arm-wrestling, all caused non-pathological shaft fractures. Pathological fractures usually affect the proximal humerus but fractures through tumour deposits (particularly in multiple myeloma) and in patients with severe osteoporosis can be seen in the shaft.

The humeral shaft connects the polyaxial shoulder joint with the elbow, which acts essentially as a hinge. The large range of motion in all planes at the shoulder joint means that malunion of shaft fractures are, for the most part, well tolerated. The humerus is well covered with muscle, minimizing the aesthetic impact of mild degrees of malunion.

The lateral aspect of the shoulder is covered by the deltoid muscle, innervated by the axillary nerve (a division of the posterior cord of the brachial plexus), which runs around the humeral surgical neck along with the circumflex humeral artery after passing through the quadrilateral space.

The anterior (or flexor) compartment of the arm contains the coracobrachialis, brachialis and biceps muscles. These are innervated by the musculocutaneous nerve, which runs in the plane between biceps and brachialis and has as its terminal division the lateral cutaneous nerve of the forearm. Brachialis receives an additional innervation from the radial nerve, the nerve that primarily supplies the posterior (or extensor) compartment muscle, triceps. The radial nerve arises from the posterior cord of the brachial plexus, runs along the posterior surface of the humerus in the spiral groove between the lateral and medial heads of triceps before passing through the lateral intermuscular septum to pass anterior the elbow joint, supplying the radial wrist extensors (extensor carpi radialis brevis from its posterior interosseous division) and brachioradialis.

The median (Figure 12.35.1) (from the lateral cord of the brachial plexus) and ulnar nerves (from the medial cord) traverse the arm without normally innervating any structures, the median nerve crossing anterior to the brachial artery from lateral to lie medial in the antecubital fossa, and the ulnar nerve passing from the anterior to the posterior compartment through the medial intermuscular septum, crossing the elbow behind the medial epicondyle within the cubital tunnel.

The differential innervation of the compartments and sites of the neurovascular bundles dictates the common approaches to the humerus. The proximal shaft is usually approached through an anterolateral approach, which is extensile proximally into the deltopectoral approach to the shoulder and distally into the anterior approach to the elbow and forearm (Henry’s approach). This uses the plane between the territories of the axillary and radial nerves posteriorly and the musculocuta-neous nerve anteriorly; brachialis is split to expose the anterior surface of the humerus, relying on its dual innervation, giving access to the whole length of the bone. The patient is positioned supine with the arm on a radiolucent hand table. The resultant scar is obvious.

Distal shaft fractures, particularly with extensions into the distal metaphysis, are generally exposed through a posterior triceps-splitting or reflecting approach, which also allows exploration of the radial nerve in the spiral groove in the event of a radial nerve palsy. An early decision whether triceps continuity with the ulna, with its insertion into the olecranon, can be preserved (triceps reflection or splitting) or the extensor mechanism should be reflected in its entirety (olecranon osteotomy) needs to be made; this decision will largely be determined by the presence of a complex articular injury, requiring direct visualization of the distal humeral articular surface (and so requiring an olecranon osteotomy). Patient positioning is more difficult for posterior approaches, with the patient either lying prone or lateral with the arm in a gutter. Access for intraoperative imaging is also more challenging.

The medullary canal of the humerus runs from the head to just above the olecranon fossa distally. Access for intramedullary fixation techniques can be achieved antegrade through the humeral head (via a deltoid-splitting approach, but requiring violation the supraspinatus element of the rotator cuff). Care must be taken to avoid injury to the axillary nerve if the deltoid is split more than 4cm below the acromion. Retrograde insertion through top of the olecranon fossa is an alternative (via a posterior triceps-splitting approach) but risks creating a supracondylar fracture and makes locking in the proximal humerus more difficult.

The energy of the injury is important to determine, to gauge the likelihood of associated injuries and of a pathological fracture, both of which may influence the chosen method of treatment. An assessment of the anticipated level of compliance with rehabilitation advice is similarly influential.

Examination of the patient should exclude associated injuries, treatment of which may well take clinical priority, and the patient’s fitness for anaesthesia. Local examination focuses on the integrity of the soft tissue envelope and confirming the distal neurovascular function, in particular of the radial nerve (tested best by confirming finger extension at the metacarpophalangeal joint (MCP) joint, thumb retropulsion (an extensor pollicis longus function), and wrist dorsiflexion).

Plain film imaging in two planes, without rotating the arm, will suffice in the majority of cases. Pathological fractures will usually be in patients with known metastatic disease, but staging investigations will be needed if the primary tumour is unknown. (see Chapter 12.18). Diaphyseal fractures with articular extensions may require cross-sectional imaging if the articular element cannot be accurately defined on the plain films.

Neurophysiological investigation of nerve lesions is appropriate at between 3–6 weeks if there is no evidence of clinical recovery. Leaving studies for a few weeks will allow denervation changes to be seen.

Given that mild malunions are accommodated with good function and that union with non-operative treatment is the norm, the overwhelming majority of humeral diaphyseal fractures can and are treated non-operatively. This should be an active process, with care directed to the type of brace used and to ensure that the risk of stiffness of the adjacent elbow and shoulder is minimized. Initial treatment with a plaster splint like a Bohler slab (to cover the shoulder like a cowl and splint the length of the humerus) to control pain is indicated, with collar and cuff support applied and instruction given to keep the elbow dependent so that gravity will assist reduction. If the humeral alignment is slow to achieve, reinforcement of the advice to keep the elbow dependent and application of a forearm cast to provide a little more gravity-assisted traction is appropriate. (Non-operative management is also dealt with in Chapter 12.11.)

Once the fracture swelling and discomfort has subsided, the protective slab can be replaced with a cooptation brace for the arm. This should be kept tight so that the hydrostatic tension within the muscle compartments will support the fracture, allowing micromotion to encourage union, but not too much motion so that union is prevented. Elbow motion should be encouraged.

Union should be expected by 8 weeks, and is achieved in over 90% of cases. Functional brace use may be associated with impaired shoulder function, but gives better elbow function than use of a U-slab throughout the treatment period.

Open fractures require debridement and stabilization; in all but the most extreme soft tissue injuries, or where vascular repair is necessary, internal fixation is preferable. Patients with ipsilateral humeral shaft and forearm fractures also require stabilization of both skeletal injuries. Patients with pathological fractures are generally best treated operatively to give early pain control and return of function.

Other indications for acute fixation are relative, but include multiply injured patients (for ease of nursing care), patients where compliance with rehabilitation is likely to be poor, patients in whom a radial nerve palsy has developed during treatment (see later), and certain fracture configurations (relatively transverse fractures, especially in the proximal third, and certain distal shaft fractures when varus mal-union is less well tolerated).

Open plate fixation of fractures when operative treatment is deemed necessary remains the most common treatment modality. The approach will usually be anterolateral, although fractures of the distal third or those with associated radial nerve lesions or extension into the elbow are better treated through a posterior approach. A large fragment plate (4.5mm) should be used due to the large forces applied to the implant during healing and rehabilitation, with at least four used holes in both the proximal and distal fragments. Some authors recommend the use of supplementary prophylactic bone-graft at the initial fixation to maximize the rate and likelihood of union.

Minimally invasive plating techniques, often using locking plate systems, have been reported with high union rates. As the neurovascular structures are not exposed during the approaches, care must be taken if undertaking this type of fixation.

Initial enthusiasm for intramedullary techniques was dampened by reports of cuff impairment with antegrade insertion techniques, supracondylar humeral fracture with retrograde insertion techniques, comminution during nail insertion, fracture non-union (possibly related to poor rotational control), and problems with the instrumentation. Many of these problems have been reduced by improvements in the design and insertion techniques and instrumentation systems, and high union rates of close to 100% have been reported in series of both open and closed fractures. Similar union rates and function but more shoulder pain are reported in series that compared treatment by plate fixation or intramedullary nailing.

Intramedullary nailing has advantages in the treatment of pathological fractures (or impending fracture), where all but the most distal portion of the humerus can be stabilized, and potentially in obese patients, in whom access for open treatment may be difficult.

While less commonly used than in other bones, external fixation is used for open injuries, to bridge a ‘floating elbow’ (associated ipsilateral humeral and forearm fractures) and in damage control situations with multiple injured patients.

While injuries to neurovascular structures have been described complicating humeral shaft fractures, most series report an incidence of radial nerve palsy of 11.8%. Radial nerve palsy is significantly more common with transverse and spiral fractures. Spontaneous recovery is seen in 71% of cases and overall recovery in 88% and is little affected by management.

During the period of expectant treatment and recovery, care must be taken to ensure wrist and digit contractures do not occur that will impair the ultimate function. Provision of a splint to hold the wrist dorsiflexed in a functional position and instruction in exercises to maintain passive motion of the wrist and digit joints is essential. Splints that provide a passive extension force to the MCP joints while allowing active flexion can be used, but these splints risk creating a loss of flexion at the MCP joints if full flexion is not maintained (actively or passively) throughout the recovery period. If neurophysiological investigations suggest that the nerve has been divided, exploration and nerve grafting should be undertaken promptly to maximize the likelihood of the best possible outcome. If a long time has elapsed since the injury without evidence of recovery, tendon transfer will give a better functional outcome, although full passive motion of the wrist and digit joints is essential (see Chapter 12.25).

While most experts accept that initial expectant treatment of a radial nerve present from the time of injury is appropriate, the optimal treatment when nerve palsy arises during the treatment of a humeral shaft fracture is more controversial. Exploration of the radial nerve is advisable following palsy onset after intramedullary nailing or minimally-invasive plate fixation due to the potential for intraoperative nerve laceration. Nerve injury should be noted if a posterior approach has been used for open plating, but if an anterolateral approach has been used, exploration should be undertaken. Radial nerve palsy during non-operative treatment is the most controversial subject as in the majority of cases, exploration of the radial nerve in these circumstances showed an in-continuity lesion (as is the case with radial nerve palsy present from the moment of injury). As with radial nerve palsy from the injury, an initial period of expectant treatment and active monitoring with neurophysiological investigation at 4–6 weeks is reasonable, although early exploration can be justified.

Rates of non-union with non-operative treatment range from 2–39%. Factors associated with failure to achieve union by non-operative means include the site of the fracture, with fractures of the proximal third being less likely to unite, and the pattern of the fracture, with transverse and short oblique fractures reported as being possibly less likely to unite.

Non-steroidal anti-inflammatory (NSAID) use has also been noted to be associated with non-union of humeral shaft fractures with a relative risk of non-union of 3.7 (95% confidence interval 2.4–5.6). However, in this retrospective study, NSAID use was only significantly associated with non-union at 61–90 days following injury (with opioid use in the same time window also showing a significant association with non-union), raising the possibility that the drugs were taken because of the pain of the impending non-union, and that the drug use may not be causal of the non-union.

Non-union may be treated by open plating with onlay bone grafting or by reamed intramedullary nailing.

Humeral shaft fractures can occur at any age, although the incidence increases with age. The majority can and should be treated non-operatively as the outcome is generally good and the degree of malunion that commonly occurs is well tolerated with good function. Exceptions to the non-operative rule include open fractures, pathological fractures and humeral shaft fractures in multiply injured patients.

Non-union should occur in less than 10% of cases with non-operative treatment and can be treated with stabilization and bone grafting. While radial nerve palsy complicates over 10% of humeral shaft fractures, the overwhelming majority of lesions are in-continuity and can be treated expectantly with splintage and mobilization of the wrist and hand to maintain function, and clinical and neurophysiological monitoring to look for signs of early recovery. Prompt surgical treatment should be offered to patients in whom there is concern about radial nerve division.

When surgical stabilization is indicated, open plating and minimally invasive treatment with intramedullary nailing both give good results in terms of union and function, although there is a greater incidence of shoulder pain and dysfunction with antegrade intramedullary nailing. Minimally-invasive plate fixation has been reported, but the theoretical risk of iatrogenic nerve injury means caution should be observed. The role of external fixation in acute treatment is limited.