Plate fixation following ORIF is considered to be the gold standard when it comes to surgical treatment of diaphyseal humeral fractures [28]. The fracture can be visualised using different approaches, and a plate with screws is used. Advantages include direct visualisation of the fracture site and a high probability of anatomic reduction, with absolute stability and direct bone healing. Disadvantages include extensive dissection, iatrogenic injuries, infection, and the possibility of further operation. Van de Wall et al. demonstrated a low complication rate in a study of 102 patients treated with plate fixation [28].

There are several options for the surgical approach, and the choice is dependent on surgeon preference, patient factors, radial nerve, and vascular integrity [7, 30]. The anterolateral approach is common, especially in more proximal fractures. The radial nerve is identified and protected through the entire length of the incision. In the medial approach, the nerve is avoided, and the brachial artery identified, which can be helpful in cases with concomitant vascular injuries [7]. Lastly, in the posterior approach, variations exist in terms of triceps splitting or sparing. The radial nerve will be in close proximity, and careful dissection is required [7, 31]. Iatrogenic radial nerve palsy occurs at similar rates irrespective of which approach is used [30]. In a single-surgeon case series of 66 cases, the posterior triceps-sparing approach showed a high union rate and low incidence of radial nerve palsy (3%) [31].

A biomechanical study comparing ten-hole locking compression plates in sawbones was conducted to identify mechanical properties of anteromedial, anterolateral, and posterior fixations. The anteromedial plating was superior to the rest in all mechanical tests, except four-point bending. Although this study favours anteromedial fixation, an anteromedial approach is not recommended in patients presenting with concurrent radial nerve palsy [32].

Intramedullary nailing (IM nail) for diaphyseal humeral fractures gained popularity due to the minimal dissection needed and the preservation of the fracture haematoma, utilising relative stability and indirect healing. The disadvantages of IM nails are a higher rate of shoulder pain and re-operation [20], as well as an increased risk of rotational malignment.

Anterograde IM nails have an entry point at the greater tuberosity or apex of the humeral head, whereas retrograde IM nails have their entry point at the midline posterior triceps-splitting area. The retrograde approach is not as popular due to the specific risk of distal humerus fracture on insertion or extraction of this type of nail [7].

IM nailing is associated with iatrogenic rotator cuff injury, which can lead to shoulder pain and stiffness. As the IM nail design and technique evolves, the above symptoms have become less prominent [33, 34]. In particular, an entry point slightly lateral to the humeral shaft axis is recommended [34].

Fan et al., in a study comparing ORIF and IM nail in 60 patients, demonstrated that IM nailing had a lower intraoperative blood loss, decreased operative time, decreased hospital stay, and less incidence of serious complications, whereas union rates and functional outcomes were similar in both groups [35].

The MIPO technique is a relatively novel concept in which the aim is to achieve relative stability and indirect bone healing. The plate is placed anteriorly to reduce the risk of iatrogenic radial nerve injury in bridging mode using a minimal surgical approach. Benefits include minimal dissection, a decrease in conventional plating complications, and avoiding any shoulder complications that can arise when an intramedullary nail is used [36]. Although the MIPO technique is quite promising, the fact that a direct reduction of the fracture is not achieved leads to a possible rotational difference between the two sides. In a small study, this has been shown to be of little clinical significance, as patients have been found to have good or excellent functional results [37].

The use of the external fixation (Ex-Fix) technique is quite limited. It is usually a temporising option in the poly traumatised patients requiring stability with minimal systemic insult and when the soft tissues do not allow internal fixation, but they can also be used in severely comminuted fractures, open fractures, or infection cases. Bicortical pins are inserted at the lateral aspect of the humerus, taking care to avoid a neurovascular injury on insertion. Rods are used to connect the pins with each other. Particular care should be taken to achieve length and avoid mal-rotation. The average time to union with an Ex-Fix is fourteen weeks [7, 20].

The most common complication of humeral shaft fractures is radial nerve palsy [38]. A primary palsy is caused by the initial injury whilst a secondary palsy can be caused by fracture callus or scar tissue compromising the nerve during conservative treatment. In iatrogenic injury, iatrogenic secondary palsies can be a result of closed management with or without fracture manipulation or surgical intervention [39].

This manifests as paraesthesia of the dorsum of the hand with weakness of thumb abduction, finger and thumb extension at the MCPJs, and dorsiflexion of the wrist [7]. The absence of brachioradialis or extensor carpi radialis longus could also help identify a more proximal lesion. About 50-68% are reported to present as complete palsies [40-42]. Chang and Ilyas recommend that radial nerve palsies with humerus fractures can be classified as type 1 neuropraxia, type 2 incarcerated, type 3 partial transection, and type 4 complete transaction [43].

The incidence of primary radial nerve palsy with closed humeral shaft fractures ranges from 2% to 19% [30, 41, 42, 44-48]. Two literature reviews on the subject suggest the average incidence is 11.8% or 16.3% [49, 50].

Both primary and secondary radial nerve palsies are associated with fractures of the middle third or at the junction of the middle and distal third [7, 30]. A cadaveric study identified that the radial nerve was at risk in the posterior midshaft region at the distal aspect of the deltoid tuberosity and the lateral distal third of the humerus around 11cm proximal to the lateral epicondyle [51]. There are reports suggesting a higher incidence in either region, with another finding equal incidence [41, 42, 52].

Operative iatrogenic injuries are common complications ranging from 0 to 43% and are more common following plate fixation (2-18%) than intramedullary nailing (0-5%). The surgical approach does not seem to significantly impact the incidence [30, 31, 35, 53-68].

Rarely, ulnar and median nerves can also be injured. These typically occur in open fractures with significant soft tissue injury. Stahl et al. reported a case of a closed humeral shaft fracture managed with intramedullary nailing where the ulnar nerve was found to be transected on subsequent exploration. It is unknown whether this was a primary or secondary injury, as no pre-operative neurological deficit was documented [69]. Streufert et al. noted iatrogenic ulnar palsy in 1.2% when plating [30].

1.6-3% of cases can be associated with a partial or total plexus injury. As such, a thorough neurology exam is crucial. The presence of a combined axillary and radial nerve palsy would suggest damage to the posterior fascicle of the brachial plexus [7].

Thankfully recovery rates are quite good. Primary palsies have a spontaneous recovery rate of 70.7%, increasing to 88.1% following delayed exploration [49]. Streufert et al. reported similar findings (74%) [30].

Primary palsies that do not recover are likely to have been associated with more significant energy trauma. As a result, more primary palsy patients go on to require tendon transfers or wrist fusion (22% vs. 0%) [30].

Palsies of iatrogenic nature are usually temporary and resolve spontaneously in 70-100% of cases with a shorter recovery period than primary palsies (4.1 vs. 5.5 months) [30, 31, 42, 44, 54-65, 68, 70-76]. However, if the palsy is caused by local soft tissue compression, it might not improve and require surgical exploration.

There is a clear consensus that fractures requiring fixation (open fractures, associated vascular injuries, and fracture configurations not amenable to non-operative management) with concomitant radial nerve palsies should have early exploration of the nerve. However, the management of nerve palsy in the remainder of the injured is somewhat contentious.

Traditional teaching dictates that primary palsies should be managed expectantly and only explored if there is no evidence of recovery by 2-3 months [ 48, 76, 77]. Given most patients recover spontaneously, this would avoid risks associated with surgery. Furthermore, patients managed with early exploration compared to observation have been shown to have no difference in outcomes [43, 49, 78].

Recent studies have, however, made arguments to the contrary, as follows:

• Late exploration prolongs disability, rehabilitation and could compromise nerve recovery, particularly if delayed by 12 months and possibly from as early as 5 months [40, 43, 48, 79].
• A significant number of primary palsies do not recover spontaneously (approximately 30%) [49].
• Early exploration allows characterisation and classification of the injury to guide treatment, inform prognosis, and provide peace of mind [43].
• Fracture stabilisation can prevent late nerve injury or incarceration. Furthermore, a repaired nerve will benefit from a better healing environment (less traction, motion, or callus formation to impede nerve regeneration) [43].
• Delayed repair can lead to increased muscular atrophy and motor endplate loss compromising recovery and thus a significant interval loss of patient function and livelihood [43].
  • Early surgical repair is technically safer and easier due to the lack of scarring, nerve retraction & risk of nerve entrapment in the callus. Furthermore, fixation would facilitate primary bone healing and decrease the chances of nerve entrapment or compression in the callus [43].
  • Yesil et al. describe a case where the sharp tip of the fracture had penetrated into the nerve dividing the nerve into two bands which would have been unlikely to recover with conservative management and could have been damaged further in case of fracture site movement or encasement in callus [80]. Another case required sharp dissection of the nerve at the bone for repair due to callus encasement [81].

Electrodiagnostic studies can be used to monitor recovery after 3 weeks and repeated after 8 weeks if no clinical improvement [40, 82]. Ongoing work suggests diffusion-tensor MRI could distinguish between neurotmetic and axonotmetic injury [83, 84].

With regard to secondary palsies, recent studies have demonstrated that they can be managed similarly to primary palsies [30, 39, 49, 68].

Closed humeral fractures were always believed to have a high union rate. Multiple studies have reported >90% non-operative union rates [15-17, 21, 22, 29, 38, 85]. Recently, however, some studies have reported higher non-union rates of 13-23% [18, 20, 24, 86], and a large multicentre study published in 2020 showed an even higher rate of 29% in North American hospitals [26] as previously discussed.

Aside from the common systemic factors associated with non-unions (e.g., advanced age, smoking, alcoholism, and nonsteroidal anti-inflammatories), psychiatric pathology (psychotic/bipolar disorder, dementia, or multiple involuntary psychiatric admissions), long oblique and spiral fracture patterns, as well as, proximal-third fractures and distal-third fractures are also associated with non-union when managed conservatively [25, 29, 87].

The reported incidence of post-operative non-union ranges from 2-10%, with no difference between IM nailing and plating [20, 29, 35, 53, 65, 66, 88].

An interesting score called RUSHU (Radiographic Union Score for Humeral fractures) aims to predict fractures that will not unite by evaluating the characteristics of a radiograph 6 weeks after the injury. This is pending external validation but might be a very useful tool in the future [89].

The humerus is quite forgiving when it comes to deformities; up to 20° anterior angulation or 30° varus angulation are generally well tolerated [19, 80]. With regards to rotation, a study found that an average difference of 6° internal rotation and 9° external rotation resulted in no significant difference in functional scores [37].

It is a fact of life for any surgical treatment, but especially plate fixation. Infection rates range from 2-4% [53, 65, 66]. Pidhorz et al. broke these down by fixation methods and reported an average of 4% for plating, 1.6% for IM nails, and 4% for external fixation [7]. In contrast infection is rarely a problem in non-operative management (0%) [20].

Vascular injuries are rare (0.5-3%) and are mainly due to brachial artery rupture. These require urgent support from the vascular surgeons since a provisional shunt might be required, which allows the fracture to be stabilised before any blood vessels can be definitively repaired [7].

Antegrade nailing is associated with shoulder pain and a worse range of motion [54]. 56% experienced pain at the shoulder or fracture site, and 41% had poor shoulder function [73]. Shoulder function impairment can result from a violation of the rotator cuff, tendon injury, impingement (17%), and adhesive capsulitis [35, 54, 90]. In a cadaveric study, Schwarz et al. found an incidence of iatrogenic infraspinatus tendon injury in 17.5% [34].

Muccioli et al. found the incidence of supraspinatus tendon lesions to be 12.5% in his study of 40 antegrade humeral nails, with 2 attributed to the high nail positioning. All, however, were asymptomatic. Supraspinatus tendon lesions are likely to be of little significance given the prevalence in asymptomatic healthy individuals was 16% in ultrasound studies. The long head of biceps tendinopathy, however, was symptomatic and half of which were attributed to a technical error in placement of the locking screws [33].