1 Introduction

Brachial plexus injuries (BPIs) are complex, debilitating injuries, largely affecting the young adult male population, with an age range between 15 and 40 years.^1–5 The incidence of BPI is estimated at around 1.64 per 100,000 people annually.^2,3,6 The incidence has increased over the past 50 years and is attributed to advancing technology, industry, popularity of extreme sports, and the improved survival of patients following major trauma.^2,7 The management of BPI has concurrently improved with advances in special imaging and microsurgical techniques.⁷

Internationally, closed traction injuries from motor vehicle accidents (MVAs) are the most common cause of adult traumatic BPIs (TBPIs) and constitute 95 % of cases.^2–5,8 The proportion of traction injuries is lower, and the proportion of penetrating injuries is higher, in South Africa, an upper-middle-income country with a high trauma burden, compared to international cohorts.^9–11 In South Africa, 1995, Boome found that 40 % of patients sustained stab injuries and 7 % had gunshot wounds, representing a total 47 % TBPIs as a result of penetrating trauma.⁹ Trauma is estimated to account for 25 % of the presentations to emergency departments in the KwaZulu-Natal province.¹⁰ Madsen et al. performed an audit on penetrating neck injuries in KwaZulu-Natal.¹² Of the 452 patients who sustained stabs and the 58 with gunshots, 17 (4 %) and nine (16 %) had BPIs respectively.¹²

In addition to the mechanism; time from injury, number of associated injuries, and level of involvement, all influence the clinical presentation, management and outcomes for patients with TBPIs.^1,7,13,14 When associated upper torso trauma and head injury are present there may be delays to identification of BPI.^2,13 Surgical management is dependent upon the mechanism, structures injured, local soft tissues and chronicity of injury (due the degeneration which occurs at the motor endplate at 20–24 months of absent innervation).^5,15 The priorities of surgery are to restore elbow flexion, followed by the stabilisation of the shoulder, then shoulder abduction, shoulder external rotation, and hand sensibility, in that order.^1,5,6,14,16 The functional outcomes depend upon mechanism, timing and type of surgery, level of plexus involved (lower root involvement is associated with a poor prognosis) and age (the younger the patient the better the outcome).^14,15,17,18 Kim et al. performed primary nerve repairs for patients with penetrating lacerations of the brachial plexus operated within 72 h of injury. 81 % of these patients had a Medical Research Council (MRC) grade three or better recovery to the affected brachial plexus elements.¹⁹ Nerve grafting was performed for patients with delayed presentations (operated more than 72 h after the injury) and the recovery to MRC grade of three or more decreased to 53 %.¹⁹ For nerve transfers, Nagano et al. reported an MRC recovery to grade three or more for elbow flexion in 80 % of their patients aged between 16 and 40 years who had intercostal nerve transfers for biceps function within six months of injury.^3,20 Age has an impact on the outcomes as well, patients over 40 years tend to have slower recoveries and less motor function when compared to those around the age of 20.¹⁵

There is a paucity of data relating to the outcomes of surgically managed traumatic adult BPIs within the South Africa context. This retrospective review aimed to determine the demographic characteristics, mechanisms of injury, and management outcomes of adult traumatic BPIs managed surgically.

2 Methods

We conducted a retrospective chart review of all patients who were 18 years of age and older, that underwent surgery for the management of a TBPI, at a single training hospital between the 1^st of July 2011 and the 30^th of June 2021. The ‘Strengthening the Reporting of Observational Studies in Epidemiology’ (STROBE) guidelines for cross-sectional studies were used for reporting.²¹ Eligible patients were identified through an electronic search performed on the digitised health records of the hospital (Meditech – Copyright © 2022 Medical Information Technology, Inc.), where the term ‘brachial plexus’ was flagged in the operative indication or in the surgical procedure. The patients' electronic health records were then accessed to confirm inclusion, and data extracted. Records of patients aged 18 years or older at the time of surgery, who underwent surgery of the brachial plexus or a procedure to manage the sequelae of TBPI, with a minimum follow-up duration of one year were eligible for inclusion. Duplicate surgical records for patients undergoing secondary procedures were excluded and the information combined to represent the individual patient with multiple procedures. Patients were also excluded if their records did not include functional outcome assessment more than one year after their primary procedure despite follow-up longer than one year, these records were deemed incomplete.

From the records the patients’ age, sex and injury details were captured. Injury characteristics included date of injury, referral hospital, date of initial assessment, side involved, associated injuries and mechanism of injury. Both motorcycle and car accidents were classified as MVAs causing traction injury. Pedestrian vehicle accidents (PVAs) were classified in a separate group as both blunt trauma and traction may have occurred. Blunt trauma was used to denote direct trauma to the neck and shoulder from an assault with a heavy object. MRC power grading was performed for the functions of each brachial plexus element involved including, shoulder external rotation, shoulder flexion, elbow flexion, elbow extension, wrist extension, wrist flexion, extrinsic finger flexion and extension, and intrinsic finger flexion and extension. These grades were captured for the affected brachial plexus elements, at both the initial presentation and most recent follow-up for each patient. It was noted whether the patient had undergone special investigation with electromyography (EMG), contrasted tomography (CT) myelogram, or magnetic resonance imaging (MRI). Pre-operative diagnosis included whether the injury was preganglionic and the specific brachial plexus element(s) involved (determined from the clinical findings or special investigation report).

The date, nature of and number of component procedures of the primary surgery were captured for each patient. For example, a patient may have had a primary surgery that comprised a nerve graft of the C5 root as well as a spinal accessory to suprascapular transfer (SAN-SSN) nerve transfer and an Oberlin nerve transfer. This was captured as a primary surgery comprising a graft with two nerve transfers, each of the component procedures recorded separately, and represented as a primary surgery with 3 component procedures. Subsequent second and third surgeries were also captured. The term exploration was used to denote cases where the brachial plexus was explored and no injury was identified, the term neurolysis was used if a neuroma in continuity was identified and neurolysis was performed. A single Oberlin transfer entailed transfer of the flexor carpi ulnaris fascicle of the ulnar nerve to the biceps brachii branch of the musculocutaneous nerve in the upper arm.²² With the addition of the flexor carpi radialis fascicle of the median nerve transferred to the brachialis branch of the musculocutaneous nerve, the transfer was termed a double Oberlin and this was tallied as two separate nerve transfers.²³ The Somsak procedure entailed transfer of a radial nerve branch to triceps brachii (long head or medial head) to the anterior division of the axillary nerve.^24–26 Spinal accessory nerve branch transfer to the suprascapular nerve was abbreviated to SAN-SSN.²⁶ An intercostal nerve transfer entailed transfer of three ipsilateral intercostal nerves to the musculocutaneous nerve without utilising an intermediary graft and was tallied as three nerve transfers.²⁰ If the ipsilateral phrenic nerve was used as a donor to the musculocutaneous nerve, intermediary nerve graft was utilised.²⁷

TBPI's represent a vast spectrum of disease. We managed patients according to general principles, based on the timing and deficits present, and donors available. Hill et al. provide an expanded description of all options available in their 2021 publication. The treatment strategies relevant to this cohort are described here.²⁸ Patients presenting acutely with penetrating stab injuries were booked for the next available hand surgery slot and were explored. Any haematoma was drained and neurolysis of early scar formation was performed if there was no laceration to the brachial plexus elements noted. A primary repair was performed if this could be done without tension, or nerve grafting if tension-free repair was not possible. For delayed (more than six months post injury) presentations or where local soft tissues could not provide a healthy bed for the regenerating nerve, the addition of nerve transfers were occasionally required. Patients with injuries from MVAs, PVAs and blunt trauma were managed on a case-by-case basis, utilising a combination of the time from injury, presence or absence of a supraclavicular or infraclavicular Tinel's sign, EMG studies, special imaging, and MRC grading assessment to determine if roots were available for exploration and grafting to viable donor motor end plates, or if nerve transfers were indicated. Typically, exploration and grafting were performed for blunt or traction TBPIs within six months of injury in patients with a positive but non-advancing or slowly advancing Tinel's sign, minimal or no clinical or EMG motor recovery by six months, and or imaging confirmation of roots available for grafting. In those patients not meeting these criteria, patients over 40 years, or in cases where certain elements recovered and others had not, nerve transfers were performed up to a year post injury.

The number of additional surgeries were also recorded. These included subsequent nerve transfers performed at a second setting for patients that could not have multiple nerve transfers in a single setting due to theatre unavailability or patient physiological concerns, or in patients where the recovery after nerve grafting was poor. No primary tendon transfers were performed, patients with poor recovery more than one year post injury despite previous nerve surgery were offered tendon transfers as salvage procedures to improve the affected function(s). Finally, the MRC grading at most recent follow-up and notation of whether an application had been made for a disability grant, concluded the data captured.

For statistical analysis the patients’ final functional recovery outcome was categorised into two groups, using the average MRC grade of the affected function(s). The poor function group comprised individuals with an averaged MRC recovery to less than three, an averaged MRC recovery to three or more was categorised as the good function group. Statistical analysis was performed using Jamovi statistical software (version 2.2.1). Continuous variables are reported as means (standard deviation [SD], range) or medians (interquartile range [IQR], range), and categorical variables as counts and frequencies. The Shapiro-Wilk test was used to analyse the distribution of data. Binomial logistic regression was used to measure association between independent variables and the outcome measure and were represented by odds ratios (ORs) and 95 % confidence intervals (95 % CI) where appropriate. All tests were two-sided, and the level of significance set at p < 0.05.

3 Results

The electronic search of digitised patient health records of patients that received surgery for TBPI yielded 60 patients eligible for inclusion. There were 13 exclusions. Six were duplicate records generated for patients that had more than one brachial plexus surgery, and 7 records were incomplete. There were no cases of bilateral injury.

A total of 47 patients were included in the final analysis with a median age of 32 years (IQR 8, range 18–63 years) and 41 (87.2 %) were male. Table I details the patient demographics and injury characteristics. Most patients were referred from a local regional level hospital (n = 41, 87.2 %) within a median 58 days (IQR 83, range 1–338) from the time of injury to first assessment at the specialised hand unit. The most common mechanisms of injury were MVAs followed by penetrating stab wounds, representing 23 (48.9 %) and 18 (38.3 %) cases respectively. The remainder of injuries included one penetrating gunshot wound (2.1 %), two cases of blunt trauma (4.3 %) and three PVA's (6.4 %).

Most injuries were left sided (n = 32, 68.1 %), three (6.4 %) patients had documented Horner's syndrome, 17 patients (36.2 %) had special imaging and 20 (42.6 %) underwent electromyography during their treatment. Twenty (42.6 %) patients had a documented root avulsion of at least one root and were classified as having preganglionic injuries. Upper trunk injuries were most common, recorded for 17 (36.2 %) patients, followed by pan plexus injuries in 15 (31.9 %) patients. A significant proportion of patients (n = 18, 38.3 %) had at least one associated injury, five of these patients had two associated injuries (5 of 18, 27.8 %) and two patients had four associated injuries (4 of 18, 22.2 %). The most associated injury was a tibia fracture (4 of 18, 22.2 %). A forearm fracture, scapular fracture, traumatic brain injury or cervical spine injury were the next most common injuries each accounting for three of the patients with associated injuries (3 of 18, 16.7 %). The median MRC grading of the affected brachial plexus element(s) function group of patients at their initial assessment was 0.0 (IQR 1.0, range 0.0–2.0).

Table II summarises the pertinent information regarding patients' surgical procedures. The mean time to initial surgery from date of injury was 135 days (SD 90.3, range 3–340). Forty-seven primary surgeries were performed comprised of 88 component procedures. One patient's primary surgery constituted an exploration only (1 of 47, 2.1 %), three patients received a neurolysis solely (6.4 %), grafting was the exclusive treatment in ten patients (21.3 %), two patients had a primary nerve repair only (4.3 %), the exclusive use of nerve transfers (one or more) occurred in 25 patients (53.2 %), the remaining six patients (12.8 %) had a combination of procedures. Sixty-six nerve transfers were performed in the primary setting (66 of 88, 75 %). SAN-SSN was the most common transfer (21 of 88, 23.8 %), and was performed most commonly in combination with a single Oberlin and a Somsak (a triple transfer) on six occasions (6 of 21, 28.6 %), a nerve graft on three occasions (3 of 21, 14.3 %), and an intercostal transfer on four occasions (4 of 21, 19.0 %). Six (12.7 %) patients received intercostal nerve transfers comprising 18 component procedures (three intercostals in each patient transferred to the musculocutaneous nerve), and three (6.4 %) a phrenic transfer with an intermediary graft. Fourteen (29.8 %) patients had one or more additional surgeries, 11 (23.4 %) had one additional surgery and three (6.4 %) had two additional surgeries. In the secondary setting six patients had nerve transfers (6 of 14, 42.9 %) and eight patients had tendon transfers (8 of 14, 57.1 %). The third operation was a tendon transfer for two patients (2 of 3, 67 %) and a wrist fusion in the remaining patient (1 of 3, 33 %). The patients followed up for a median 21 (IQR16.5, 12 to 105) months after their initial surgery. The median MRC grade of patients at their most recent follow-up appointment for all affected muscle groups was 2.00 (IQR 3.00, 0.00 to 4.00). Twenty-nine (61.7 %) of the patients were documented as having applied for a disability grant.

Fourteen (29.8 %) patients had a good outcome (averaged MRC of affected brachial plexus element function groups of three or more) and 33 (70.2 %) patients had a poor outcome. Table 3 compares the demographic, injury, and surgery characteristics between the two groups. With respect to demographic characteristics, the two groups were similar. Considering the diagnosis of brachial plexus element involvement, all patients diagnosed with pan plexus injuries had poor outcomes (15 of 33, 46 %). There was no difference in outcome comparing penetrating injuries to closed injuries (p = 0.386, OR 1.75, 95 % CI 0.49 to 6.20). Regarding the primary surgical procedure and its component procedures, the majority of cases where nerve grafting was performed fell in the poor outcome group (11 of 33, 33 %) as opposed to good outcomes (3 of 14, 21 %). All intercostal nerve transfers fell in the poor outcome group (18 of 60, 30 %), as did all phrenic nerve transfers (3 of 60, 5 %). In those patients who received second and third operations, the majority had poor outcomes (12 of 33, 20 %) as opposed to good outcomes (2 of 14, 14 %). There was no difference in outcome in patients that had a single surgery as compared to patients that required additional surgery (p = 0.145, OR 0.29, 95 % CI 0.06 to 1.53).

4 Discussion

This study aimed to describe the basic demographics, injury characteristics, and management outcomes, of adults with surgically managed brachial plexus injuries. In keeping with previous investigation our cohort was likewise comprised of predominantly young male patients. Eighty-seven percent of patients were male with a median age of 32 years, consistent with previous reports ranging from 82 % to 98 % male preponderance, and mean ages between 24 and 31 years.^{1,4,6,16,29,30}

The median time from injury to the initial assessment of our cohort was 58 days. Similarly, Zhang et al. reported a median time of 2 months from injury to surgical assessment in their study which investigated reasons for delayed presentation of adult traumatic BPI.³¹ They defined delayed referral as a period of “more than 3 months from the date of injury to the date of initial evaluation by a brachial plexus surgeon”, a cutoff which seems clinically relevant provided the assessment is followed imminently with surgery.³¹ The mean time to surgery in our series (135 days or 4.8 months) fell within the broad range of that from previous investigations, from 3.6 to 16.8 months.^1,29,30 The outlier to this group (with average time to surgery of nearly 17 months) was interestingly from another upper-middle-income country, Indonesia.^11,30 The authors suggested that the archipelago geography of the country, with its associated difficulties in travelling between them, may have contributed to the physical delays to presentation to their trauma centre.³⁰

A systematic review by Martin et al. determined that 90 % of patients operated within 3 months of BPI regain an MRC grade of three or more, which drops to 36 % if operated after 12 months.¹⁶ For traction injuries specifically, early referral to brachial plexus surgeons allows for timeous assessment and monitoring of spontaneous recovery, failing which, surgery can and should be expedited for maximal recovery potential. Zhang et al. found in their cohort that referral was delayed due to late referral by the local clinics or hospital, or misdirected referrals.³¹ Other causes mentioned were management of associated injuries, delays to obtain electrodiagnostic testing and patients seeking different opinions.³¹ Whilst the reasons for delay were not specifically investigated in our study, it was notable that a large proportion of the patients (n = 18, 38.3 %) had associated injuries which may have resulted in a delay to diagnosis. Furthermore, as with Zhang's study, the hospital study site services a geographically large area, with reduced access to health care, and delays to transfer for urgent surgical treatment.³² Suroto et al. cited the geography, socioeconomic and educational systems of the islands making up Indonesia as the main reasons for the long delays.³⁰

In international literature, MVAs are the commonest cause of adult TBPIs.^{1,3,4,6,8,13,15,31,33}Motorcycle accidents account for the majority of these injuries, reported as upwards of 90 % in some series.^1,8,15,29,30 Motorcycles were identified as the main mode of transport to workplaces, and this is thought to be the reason why there are high numbers of these accidents.³⁰ Only 48.9 % of the patients had BPI sustained from MVAs (including motorcycle accidents) in our cohort, whereas a large proportion (38.3 %) had injuries from penetrating stab wounds. The elevated number of stab wounds in our study may be due to the increased levels of interpersonal violence in South Africa. Most violence in South Africa is interpersonal, with homicide rates standardised for age seven times higher compared to the averages in other regions of the world.^34,35 In 2013, 85 % of the deaths from interpersonal violence in South Africa were as a result of penetrating stab injuries compared to the global figure of 24 %.³⁶ Mitton et al. noted a large number of these fatalities associated with stabs to the chest and neck area in their retrospective audit at a medico-legal laboratory.³⁶ Though these figures are alarming, injuries sustained from penetrating stabs have been found in previous work to have better outcomes. Our sample was underpowered to detect an association with outcome comparing mechanism of injury, however we did note that 50 % of the TBPI's in the good outcome group had a penetrating stab or gunshot mechanism.

Pan plexus injuries account for between 39 and 50 % of cases globally, and upper trunk injuries between 23 and 60 % of cases.^1,13,31 This was not dissimilar to our study, where upper trunk injuries comprised 36.2 % of the patients and 31.9 % had pan plexal involvement. The level of plexal involvement has established implications in the determination of prognosis.¹³ We found that all of the patients with pan plexus injuries had a poor outcome. A recent systematic review by Martin et al. supported the findings of previous TBPI investigation as well as our findings.¹⁶ Martin et al. found that good results occur in patients with C5 – C7 TBPIs, and a further improvement in outcome for isolated C5 – C6 injuries.¹⁴ The outcomes were poor (MRC of two or less) for pan plexal injuries or in those with established preganglionic injuries.¹⁶

With respect to the nature of surgery performed, Martin et al. noted better results with intra-plexal donors for nerve transfer, postulating that this could be due to a higher axonal count of intra-plexal donor nerves.¹⁶ We found that all intercostal and phrenic nerve transfers (extra-plexal donor nerves) had poor outcomes. In addition to the axonal contribution of the donor intra-plexal nerves it should be noted that extra-plexal transfers typically have a greater distance from donor to recipient for reinnervation, often require an intermediary graft, and are commonly performed for devastating pan-plexal injuries in which only extra-plexal transfers are available. Each of these factors could contribute to poor outcomes and confound comparison.

Considering that poor outcomes occurred in 70.2 % of patients in our cohort, it is reasonable that 61.7 % of patients were recorded as having made application for a disability grant. This highlights the social burden TBPI places on a society, with increasing unemployment rates and a large percentage of people dependent on state support grants for social security.³⁷

Due to the retrospective collection of data, inadequacy of records, and number of patients lost to follow up, the resultant sample size was small. Furthermore, the spectrum of TBPI was heterogenous in mechanism, diagnosis and treatment. Consequently, the analysis of association between outcomes groups was underpowered to detect significance for several clinically relevant groups. Future investigation of a prospective nature, investigating specific subgroups of injuries, is likely to expose factors that affect functional outcome.

5 Conclusion

Adult traumatic BPIs in this South African sample typically presented more than two months after injury and were comprised of a high proportion of penetrating injuries. Just under a third of surgically managed patients had a good outcome. Pan plexal injuries have uniformly poor outcomes. We recommend early referral for all traumatic BPI to a unit that manages BPI to improve outcomes.

Funding statement

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethical statement and institutional ethical committee approval

The authors declare that this submission is in accordance with the principles laid down by the Responsible Research Publication Position Statements as developed at the 2nd World Conference on Research Integrity in Singapore, 2010 and the Code of Ethics of the World Medical Association (Declaration of Helsinki). The study complied with the South African Department of Health ethics guidelines (2015), and the University of KwaZulu Natal policy on research ethics. Prior to commencement of this research the appropriate ethical approval was obtained from the Biomedical Research Ethics Committee of UKZN (BREC/0004533/2022).

Guardian/patient consent

Not applicable for this retrospective chart review.

CRediT authorship contribution statement

Henry Lunga: Conceptualization, Data curation, Writing – original draft, Writing – review & editing. Megan O'Connor: Methodology, Data curation, Formal analysis, Writing – original draft, Writing – review & editing, Supervision. Antoine GL. Rocher: Data curation, Writing – review & editing, Supervision. Leonard C. Marais: Conceptualization, Methodology, Writing – review & editing, Supervision.