Peripheral Nerve Injury After Upper-Extremity Surgery Performed Under Regional Anesthesia: A Systematic Review.

Purpose: Peripheral nerve injury (PNI) is a known adverse event following upper-limb surgery performed under brachial plexus regional anesthesia (RA). When PNI is noted after surgery, patients and providers often have questions about which factors might have contributed to this complication. This systematic review evaluates the literature on hand and shoulder surgeries performed under ultrasound-guided, plexus RA to identify factors potentially associated with PNI, including the surgery location and block type. We hypothesized that shoulder surgery might be associated with an increased risk of PNI compared to hand surgery.
Methods: A systematic review of the relevant literature was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Only prospective studies on the use of ultrasound-guided, preoperative, brachial plexus RA for hand or shoulder surgery on adult patients were included. Study groups were categorized according to surgery location and block type and compared across a number of factors via univariate and multivariate analyses.
Results: A total of 3,037 abstracts were screened; 192 full-text articles were independently reviewed by 2 of the authors; and 53 studies were included in the systematic review analysis. Following hand surgery, PNI was reported at an average rate of 1.35% ± 3.21% across 836 subjects in 40 study groups; after shoulder surgery, the average rate was 0.50% ± 1.57% across 3,383 subjects in 15 study groups. There was no statistically significant correlation between the incidence of PNI and surgery location (P =.70) or any of the most common approaches for brachial plexus anesthesia in the multivariate analysis. Conclusions: This systematic review of over 50 articles on upper-limb surgery performed under RA shows no association between the incidence of PNI and the location of surgery or type of brachial plexus block. Type of study/level of evidence: Diagnostic II.

Regional anesthesia (RA) of the brachial plexus is a valuable technique in upper-extremity surgery. Compared to general anesthesia, RA can improve postoperative pain, speed rehabilitation, and reduce the incidence of postoperative nausea and vomiting following both hand and shoulder surgery.1, 2, 3, 4 However, RA carries a risk of iatrogenic nerve injury, a rare but potentially devastating complication for affected patients., Despite several studies on the topic, the exact incidence of nerve injury caused by plexus RA remains unclear, with estimates ranging from 0.01% to 14%.7, 8, 9, 10

Surgical nerve injury is a dreaded postoperative complication occurring either secondary to direct intraoperative injury or, more commonly, as an indirect result of traction on or contusion to the nerve. Certain factors increase the risk of surgical nerve injury, such as patient positioning, duration of the surgery, retraction technique, and anatomy of the surgical field.10, 11, 12, 13

Whether caused by needle-nerve contact during RA administration or by surgical trauma itself, postoperative peripheral nerve injury (PNI) can range from neurapraxia, which resolves spontaneously over the course of a few months, to complete nerve transection or neurotmesis. When PNI occurs following surgery performed under RA, anesthesiologists, and surgeons often try to discern whether the complication is most likely a result of the surgery itself or the block. This quandary was studied recently by Droog et al in their trial of 297 patients, which did not identify any specific risk factors for nerve injury following distal upper-extremity surgery performed under RA.

A large body of data on iatrogenic nerve complications exists in the surgery and anesthesiology literature on upper-extremity surgery performed under brachial plexus RA. The studies, which report nerve complications as a nonprimary end point, present an opportunity to more robustly answer the following question: can PNI be reliably associated with certain types of upper-limb surgery or block procedures? Given our clinical experience, we hypothesized that the location of surgery (shoulder surgery, in particular) would be associated with an increased risk of PNI. In this systematic review, we evaluated the literature on hand and shoulder surgeries performed under ultrasound-guided plexus RA to identify factors potentially associated with PNI, including the surgery location and block type.

Materials and Methods

A systematic review of the relevant literature was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Fig.).

Figure

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram for systematic reviews. From Page et al. For more information, visit http://www.prisma-statement.org/. US, ultrasound.

Search strategy and eligibility criteria

PubMed, the Cochrane Library databases, EMBASE, and Scopus were searched in August 2020 for articles published between 1980 and 2020 using key words related to brachial plexus anesthesia, hand or shoulder surgeries, and nerve injuries. The references of relevant studies and prior systematic reviews were reviewed for additional studies. Duplicate and non-English articles were excluded, as were nonoriginal articles (eg, letters, reviews, meta-analyses) and abstract-only publications.

Only prospective studies (levels of evidence I and II) on the use of preoperative, brachial plexus RA for hand or shoulder surgery on adult patients were included. For the purposes of this review, “hand surgery” encompassed any upper-extremity surgery distal to the elbow, and “shoulder surgery” was strictly surgery of the shoulder joint or rotator cuff muscles. Only studies using the most common brachial plexus block approaches—interscalene, supraclavicular, infraclavicular, axillary, and suprascapular—were included. Studies with ≤25 subjects per treatment group were excluded. Articles were also excluded if they did not report postoperative nerve complications, if they did not specify the surgical site (ie, “upper-extremity surgery”) or if their primary focus was on the use of an adjuvant medication to improve block efficacy.

Only studies in which RA was performed using ultrasound guidance (vs nerve stimulation) were included, given evidence that ultrasound may improve the efficacy and reduce the complications of brachial plexus block.16, 17, 18, 19, 20, 21 However, studies were not excluded if nerve stimulation was used only as an adjunct to ultrasound guidance. Studies were excluded if RA was administered via a continuous catheter infusion (vs injected boluses), as the use of catheters in brachial plexus blockade has been associated with a higher incidence of neurologic injuries. Finally, studies were excluded if RA was performed in the emergency department or if surgery was performed outside the operating room, because of the unique challenges of those environments.

Study selection and data extraction

A total of 3,037 abstracts were screened following removal of duplicates; 192 full-text articles were independently reviewed by 2 of the authors (R.T. and C.C.). There were no disagreements on study inclusion. In total, 53 articles matched the criteria for systematic review and were included in the analysis. A third author (M.L.S.) extracted data relevant to the surgery location; block type, drug, and volume; use of adjunctive nerve stimulation; and the incidences of transient paresthesias and PNI.

Outcomes

Studies were categorized according to the level of blinding and randomization. In double-blind randomized controlled trials (RCTs), both patients and researchers assessing postoperative outcomes were blinded to the type of RA. In single-blind RCTs, only patients were blind to the block approach. In unblinded RCTs, both patients and researchers were aware of the RA method. In all RCTs, patients were randomly assigned to 1 of 2 or more intervention groups or placebo; in cohort studies, no randomization occurred.

For the analysis, each study group with a certain [surgery location] × [block approach] combination was considered a unique entity (ie, if a single study on hand surgery included 30 patients who underwent axillary block and a separate 30 patients treated with infraclavicular block, those subjects were analyzed as 2 separate study groups of 30 patients each). In 3 cases, within-article study groups were treated separately on the basis of whether or not an adjunctive nerve stimulator was used.23, 24, 25 This method was used to maintain as much within-group homogeneity as possible.

A nerve injury was defined in this review as a new-onset motor or sensory deficit, persistent paresthesia, allodynia, or dysesthesia in the operated limb. Nerve symptoms confined to a very distal part of the upper extremity following hand surgery were excluded as representing surgical trauma that could not realistically be attributed to the brachial plexus block procedure.

Nerve complications reported either during the block procedure or within 24 hours following surgery were further classified as transient paresthesias. Only nerve complications present ≥30 days after surgery were classified as true PNIs. Nerve symptoms that resolve in <30 days likely represent neurapraxia, or nerve dysfunction without anatomic disruption, and confer little long-term disability to patients. Conversely, symptoms persisting after 1 month are much more likely to represent axonotmesis or neurotmesis, which are disruptive nerve injuries that carry more serious, and often permanent, prognoses.,26, 27, 28, 29, 30

Statistical analysis

A univariate analysis was performed to compare the hand-surgery and shoulder-surgery study groups across a number of relevant variables, including the use of epinephrine, the volume of drug used in RA, and adjunctive use of nerve stimulation. The incidences of transient paresthesias and PNI were also compared and calculated as unweighted averages of the incidences reported per study. Categorical variables were compared using the Pearson’s chi-square test; continuous variables were compared using an unpaired Student t test. A multiple linear regression model was used to identify variables that could be independently associated with the incidence of PNI. Statistical significance was defined as a 2-tailed P value of <.05, reflecting the probability of a type I error. An 80% power analysis was performed using a superiority margin of a 2% difference in the rate of PNI.

Results

A total of 53 articles were included in this systematic review: 38 hand studies and 15 shoulder studies (Table S1, available on the Journal’s website at www.jhsgo.org).18, 19, 20,,23, 24, 25,31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76 These 53 articles comprised 73 discrete, surgery-block combination study groups (54 hand and 19 shoulder groups) and 6,579 total subjects (2,875 hand and 3,704 shoulder subjects; Table 1). The large majority of studies in both groups were RCTs (94.4% hand and 78.9% shoulder RCTs); most were single-blind. Hand surgeries were performed under axillary (35.2%), infraclavicular (35.2%), or supraclavicular (29.6%) block. The majority of shoulder surgeries were performed under interscalene (73.7%) or suprascapular (15.8%) block. There was no statistically significant difference in the use of nerve stimulation (P =.42) when comparing hand- and shoulder-surgery studies, but epinephrine was used more often in hand-surgery plexus RA (59.3% vs 31.6%, respectively; P =.04), and a significantly greater volume of an RA drug was injected prior to hand surgery (32.4 ± 8.9 cc vs 26.6 ± 15.2 cc, respectively; P =.05; Table 1).

Table 1

Study Characteristics

Characteristic	Hand (n = 54)	Shoulder (n = 19)	Total (N = 73)	P Value
Unique studies	38	15	53
Total subjects	2,875	3,704	6,579
Subjects per study	53.2 (24.6)	194.9 (323.0)	90.1 (174.5)
Mean (SD)	40.0 (26.0, 122.0)	67.0 (26.0, 1,319.0)	44.0 (26.0, 1,319.0)
Median (range)	6 (11.1%)	5 (26.3%)	11 (15.1%)
Study types	36 (66.7%)	10 (52.6%)	46 (63.0%)
Double-blind RCT	9 (16.7%)	0 (0.0%)	9 (12.3%)
Single-blind RCT	3 (5.6%)	4 (21.1%)	7 (9.6%)
Unblinded RCT	19 (35.2%)	0 (0.0%)	19 (26.0%)
Cohort study	19 (35.2%)	0 (0.0%)	19 (26.0%)
Block types	0 (0.0%)	14 (73.7%)	14 (19.2%)
AX	16 (29.6%)	1 (5.3%)	17 (23.3%)
ICL	0 (0.0%)	1 (5.3%)	1 (1.4%)
ISB	0 (0.0%)	3 (15.8%)	3 (4.1%)
SCL	32 (59.3%)	6 (31.6%)	38 (52.1%)	.038
SSAX				.050
SSC	32.4 (8.9)	26.6 (15.2)	30.8 (11.1)
Epinephrine used	12 (22.2%)	6 (31.6%)	18 (24.7%)	.416
Volume injected, cc	38	15	53
Mean (SD)	287553.2 (24.6)	3704194.9 (323.0)	657990.1 (174.5)
Nerve stimulator adjunct	40.0 (26.0, 122.0)	67.0 (26.0, 1319.0)	44.0 (26.0, 1319.0)

AX, axillary; ICL, infraclavicular; ISB, interscalene; SCL, supraclavicular; SSAX, suprascapular-axillary; SSC, suprascapular.

There was no statistically significant difference in the incidences of transient paresthesias (nerve symptoms reported within 24 hours of the block) or PNI (nerve symptoms reported ≥30 days after surgery) when comparing hand versus shoulder surgery (P =.50 and.34, respectively; Table 2). Following hand surgery, PNI was reported at an average rate of 1.35% ± 3.21% across 836 subjects in 40 study groups; the average incidence after shoulder surgery was 0.50% ± 1.57% across 3,383 subjects in 15 study groups.

Table 2

Incidence of Nerve Complications

Incidence of Nerve Complications	Hand (n = 54)	Shoulder (n = 19)	Total (N = 73)	P Value
Incidence of transient paresthesias				.500
Study groups reporting	50	13	63
Subjects	2,601	2,066	4,667
Mean (SD)	0.0941 (0.1276)	0.0682 (0.1005)	0.0888 (0.1222)
Incidence of PNI				.335
Study groups reporting	40	15	55
Subjects	836	3,383	4,219
Mean (SD)	0.0135 (0.0321)	0.0050 (0.0157)	0.0112 (0.0287)

None of the variables tested in the multivariate analysis were associated with the incidence of postoperative PNI (Table 3). Most notably, the location of surgery (hand vs shoulder) had no statistically significant correlation with PNI (P =.70). The study was adequately powered, with both groups surpassing the sample size required (107 subjects) to detect a 2% difference in the rate of PNI. Similarly, the occurrence of transient paresthesias within 24 hours of the block procedure did not translate to a higher risk of PNI at ≥30 days after surgery (P =.25). Each of the 4 most common block approach types was also tested; none carried an independent association with the incidence of PNI (Table 3).

Table 3

Multiple Linear Regression Analysis

Factor	Incidence of PNI
	Coefficient	95% CI	P Value
Hand vs shoulder surgery	0.01	−0.05 to 0.07	.697
Nerve stimulator adjunct	−0.01	−0.03 to 0.01	.247
Epinephrine used	0.01	−0.01 to 0.03	.263
Volume injected	−0.00	−0.00 to 0.00	.068
Incidence of transient paresthesias	0.05	−0.04 to 0.14	.290
AX block	0.02	−0.01 to 0.04	.178
ICL block	0.02	−0.01 to 0.04	.160
ISB block	−0.02	−0.08 to 0.05	.598
SCL block	−0.02	−0.11 to 0.06	.559
Observations	50
R2 /R2 adjusted	0.173/−0.013

Multiple linear regression analysis for factors that could affect the incidence of PNI. AX, axillary; CI, confidence interval; ICL, infraclavicular; ISB, interscalene; SCL, supraclavicular.

Discussion

Following a surgical complication, patients and care providers try to understand what occurred. A PNI after an upper-limb surgery performed under RA often leads to speculation as to whether the injury occurred as a result of the brachial plexus block or surgical trauma. Although a clinical examination and/or EMG can provide diagnostic information, a definitive understanding of the etiology of PNI is often impossible.

The results of this systematic review argue against the notion that a new PNI following upper-limb surgery can be reliably ascribed to either the surgery or the block procedure. Our study revealed no statistically significant difference in the incidence of PNI following hand surgery compared to the incidence following shoulder surgery performed under brachial plexus block. Moreover, across the 53 included studies, neither the surgery location nor the RA approach was an independent predictor of PNI. While individual included studies have compared certain block approaches and found different nerve complication rates, those studies have been very small (82 subjects on average) given the low incidence of nerve complications overall (1.35% ± 3.21% following hand surgery and 0.50% ± 1.57% after shoulder surgery in our review). By aggregating the findings in over 50 studies on the topic, this systematic review offers a more complete picture of the associations between certain types of upper-extremity surgery or brachial plexus RA and PNI, or the lack thereof.

Peripheral nerve injury is a broad term that encompasses a wide range of insults to peripheral nerves. The Seddon classification system remains most commonly used, and divides PNIs into 3 major categories (neurapraxia, axonotmesis, and neurotmesis)., The mildest form of PNI, neurapraxia, is typically a result of mild compression or traction on a nerve; symptoms (eg, paresthesias, motor or sensory weakness, allodynia, dysesthesia) resolve within days or weeks and do not result in a lasting disability. The more severe types of PNI associated with RA are axonotmesis and neurotmesis, in which disruption of the nerve architecture occurs. Axonotmesis is commonly caused by stretch or crush injuries and can take months or years for a full or partial recovery., Neurotmesis is defined by complete disruption of the nerve, which will not recover without surgical intervention. In this study, nerve injuries were classified according to their long-term relevance. Paresthesias and other nerve symptoms reported during the block procedure or within 24 hours of surgery can be indicative of mild neurapraxias., Conversely, symptoms persisting at 30 days are much more likely to represent structural injuries—axonotmesis or neurotmesis—that confer real morbidity to patients and may require intervention.,26, 27, 28, 29, 30 Patients who experience a long-lasting PNI are at high risk of physical disability, emotional problems, and the development of complex regional pain syndrome, highlighting the importance of prompt evaluation of persistent postoperative nerve symptoms.,

There are several mechanisms by which RA administration can result in PNI. Direct needle-nerve contact is common during brachial plexus block. Indeed, before the widespread adoption of ultrasound, anesthesiologists often used nerve stimulators to locate branches of the brachial plexus by contacting the nerves and asking patients to report paresthesias. Even with the use of ultrasound, patients frequently report paresthesias during block procedures, and direct needle trauma is likely the etiology in some cases of PNI. Chemical neurotoxicity is another mechanism that has been proposed for RA-induced PNI. All local anesthetics are potentially neurotoxic, and some may cause both focal demyelination and axonal destruction, per animal studies.83, 84, 85 Additionally, supplemental epinephrine has been proposed to induce nerve ischemia because of vasoconstriction of supplying blood vessels, though we did not identify a correlation between the use of epinephrine and PNI in our analysis (Table 3). Finally, hematoma and bleeding are not uncommon following brachial plexus block and can cause compression within the brachial fascial compartments, resulting in ischemic changes and axonal death.,

Surgery can cause PNI through a variety of mechanisms: crushing, stretching, contusion, laceration, transection, or incorporation of a nerve in the suture repair. Shoulder surgery, in particular, occurs in a region heavily innervated by branches of the brachial plexus. The suprascapular nerve may lie within 1 cm of the glenoid rim. The axillary nerve passes 5–6 cm distal to the lateral edge of the acromion and near the deltoid split commonly used as an approach to the shoulder joint. The musculocutaneous nerve is commonly located 1–3.8 cm medial to the lower edge of the coracoid, making inadvertent nerve injury quite imaginable.,, Additionally, both of the most common positions for shoulder surgery—lateral decubitus and beach chair—have been associated with nerve injuries caused by traction or nerve compression.90, 91, 92

In hand and forearm surgery, the musculocutaneous, median, ulnar, and radial nerves are subject to similar risks of injury because of traction, contusion, and laceration. One might assume it would be easier to distinguish between nerve injuries caused by RA at the brachial plexus and those caused by these more distal surgeries. However, injuries because of brachial plexus block sometimes present in the distal distributions of its branches, mimicking more distal surgical injuries. Studies have reported nerve injuries caused by the axillary approach to the brachial plexus block presenting only in the distal distributions of the median or ulnar nerve; other studies posit that block-induced hematoma may accumulate in the medial brachial fascial compartment through which the median and ulnar nerves travel to the elbow level, producing symptoms in the distal arm or hand.,,, As a result, physicians are often unable to confidently discriminate between PNI caused by brachial plexus block and surgically induced PNI, whether surgery occurred at the shoulder or the hand.

Given the ambiguity surrounding the etiology in many cases of PNI, anesthesiologists and surgeons often attribute postoperative nerve injuries to either the block procedure or the surgery based on personal experience, anecdotal evidence, nonspecific findings on ultrasound or magnetic resonance imaging, or poorly supported notions about the safety of 1 technique over another. This review should help provide more systematic information about postoperative PNIs. We did not find that shoulder surgery was associated with a greater risk of PNI compared to hand surgery, as was our initial perception from clinical experience. Our results indicate that physicians should avoid making assumptions about the location and etiology of a postoperative PNI based solely on the surgical approach or block procedure. Instead, practitioners should be aware that the possible causes of iatrogenic PNIs are numerous and sometimes multifactorial, and should use diagnostic tools such as electrodiagnostic testing and physical examination to locate the site of nerve trauma if symptoms persist longer than a month.,

There are several limitations to our study. A meta-analysis is the preferred method for extracting conclusions from a body of literature, but this was not possible here because of the lack of standardized control or comparison groups among the included studies. The strength of this systematic review is in its aggregation of complication data from a large collection of studies comparing various approaches to the brachial plexus. This variety, however, means that few studies performed exactly the same interventions or used identical control groups, making a meta-analysis impossible. Similarly, there was heterogeneity in the way studies defined nerve complications and the time points at which patients were evaluated. To address that issue, we attempted to sort nerve symptoms into generalized categories of transient paresthesia (noted within 24 hours of surgery) and more clinically relevant PNIs (reported ≥30 days after surgery), though the authors understand this is a simplified solution. Finally, in any systematic review, some level of between-study heterogeneity is accepted to allow investigators to use the power of combined data to make more comprehensive conclusions; this review is no different.