Shoulder surgery is common and often performed in an ambulatory setting.1 Interscalene brachial plexus blockade (ISB) has long been used as a sole anesthetic or co-anesthetic in a wide array of shoulder procedures.2,3 The benefits of ISB and other regional techniques include reduced postoperative nausea and vomiting, avoidance of airway manipulation, improved postoperative analgesia, and avoidance of systemic opioids.3

Postoperative neurologic symptoms (PONS) are recognized complications of both regional anesthesia and orthopedic surgery.2,4 In shoulder surgery specifically, PONS in the operative arm can be precipitated by direct trauma from surgical dissection, ISB needle manipulation, or intraneural injection.5,6,7,8 They can also be due to ischemia from nerve traction during surgical positioning, or nerve compression from high volumes of ISB injectate or intra-articular surgical irrigation.2 Many cases behave as neuropraxias and resolve over weeks to months, but a minority persist as enduring nerve injuries.5,9,10,11 Symptoms may include persistent pain, paresthesia, sensory or motor loss in a peripheral nerve, or radicular distribution, but no formal unified definition exists.9,10

Comparative observational studies4,5,11,12 and reviews13 have reported the raw incidence of PONS in ISB vs other techniques with some suggestion of an increased risk in ISB. Nevertheless, these analyses did not account for surgical and other factors confounding block selection and technique. Randomized controlled trials (RCTs) should balance these confounding factors across study groups but previous systematic reviews of RCTs in this area3,14,15,16 have focused on analgesic endpoints, with PONS considered as a secondary outcome or not at all. By restricting this review to RCTs and quasi-randomized trials, and focusing on PONS as the primary outcome, we hoped to improve our current understanding of the role of ISB in the development of PONS. The objectives were to: 1) compare the risk of PONS from ISB (with any technique or agent) to any other regional or nonregional analgesic modality and; 2) compare the risk of PONS between different techniques or agents used for ISB.

Methods

This systematic review followed Cochrane review methodology17 and is reported according PRISMA guidelines.18 The protocol was prospectively registered on 10 February 2020 in PROSPERO (CRD42020148496).

Database and search strategy

We searched the following databases from inception to June 2020: MEDLINE (Ovid), EMBASE (Ovid), and Cochrane Central Register of Controlled trials (CENTRAL-Wiley). Abstracts from five conferences held between January 2016 and June 2020 were also hand searched (see Appendix).

Population, interventions, comparators, outcomes and study designs

We included randomized or quasi-randomized trials, reported in English, with no restriction on year of publication, and excluding editorials, letters, and notes. Nonrandomized and observational trials were excluded. We included only trials of human participants, in which at least 80% or more of the included population were undergoing shoulder surgery and were at least five years old. For the purposes of this review, we defined shoulder surgery as involving incision into and/or insertion of instruments into the shoulder (e.g., excluding closed shoulder reduction), with or without additional surgery. Trials of children less than five years old were excluded because of the difficulty in diagnosis of PONS.

The primary intervention of interest was ISB by any technique, with any composition (local anesthetic and/or nonlocal anesthetic adjuvants), of any injectate volume, with or without concomitant catheter insertion and either preoperatively or immediately postoperatively in the postanesthesia care unit. Included comparative interventions were any type of regional or nonregional analgesic technique, or ISB by a different technique, composition, or volume, including comparisons of single injection and catheter techniques.

The primary outcome was the incidence of PONS in the operative arm at longest follow up. Postoperative neurologic symptoms were included if, as defined by study authors, they satisfied each of the following conditions: 1) include one or more of paresthesia, numbness, motor loss, sensory loss, or distal arm pain in the operative arm; 2) not a composite outcome with other adverse effects not specifically related to the operative arm such as hoarse voice or dissatisfaction with care; and 3) present at least 1 week into the postoperative period. Prior to data synthesis, criterion 1 was expanded to include nonspecific neurologic symptoms and dysfunction. Secondary outcomes included 1) hoarse voice (including documented ipsilateral superior laryngeal nerve injury) and 2) dyspnea (including documented phrenic nerve palsy), where they were not part of a composite outcome and were present at least one week into the postoperative period.

Study selection, data extraction, and risk of bias assessments

Two reviewers (S. N., S. R.) independently screened abstracts in duplicate, and only those trials that did not meet the study type, population, intervention, or comparison exclusion criteria were excluded. Both reviewers then independently reviewed the full text of remaining trials in duplicate for possible inclusion. All conflicts were resolved through consensus or by a third reviewer (T. M.) as required. For included trials, a data extraction form was developed, and data independently abstracted in duplicate. The following elements were abstracted: author, year of publication, country of origin, funding sources, main inclusion criteria, main exclusion criteria, number randomized, types of surgical procedure, surgical position (lateral or sitting), elective or emergency surgery status, duration of surgery, age, sex, weight, height, body mass index, concomitant use of general anesthesia, American Society of Anesthesiologists Physical Status class, details of technique, injectate components (local anesthetic and adjuvant name and dose, adjuvant route (perineural or intravenous), total ISB injectate volume, timing (i.e., pre- or postoperatively), type of PONS (i.e., paresthesia, motor loss), duration of follow up, method of follow up (i.e., patient report by telephone assessment vs in-person physical examination), and use of a baseline neurologic examination. All conflicts were resolved through consensus, or by a third reviewer as required. All included trials were also assessed for risk of bias using the Cochrane Risk of Bias 2.0 tool.19 Assessments on each study were independently performed by two reviewers (G. L., M. M.) and compared. Disagreements were resolved using consensus or a third party (T. M.) when necessary. Additional information on risk of bias assessment is found in the Electronic Supplementary Material (ESM) eAppendix: Supplementary methods and results.

Data synthesis

Registered plans for meta-analysis were abandoned on account of high risk of bias in many trials and heterogeneity in clinical comparator groups, PONS diagnostic criteria, and the timing of PONS assessment. Data were instead synthesized and summarized descriptively.

Results

Description of included trials and study populations

From 893 unique records identified by our search strategy, 55 trials (total of 6,236 participants; median sample size, 69; range, 30–910 participants) met our inclusion criteria (Figure). It should be noted that 422 additional trials were excluded for lack of PONS report.

Figure
figure 1

Modified PRISMA flowchart

Full size image

The 55 trials were published between 1999 and 2021, with 16 trials published in 2010 or earlier. No trials reported hoarseness or dyspnea without also reporting PONS, but only eight and five trials measured these secondary outcomes, respectively. Trials and population characteristics are summarized in Table 1, ESM eTable 1, and eAppendix: Supplementary methods and results. No patients under the age of 18 were included in any of the included trials and all blocks were done preoperatively.

Table 1 Baseline characteristics of included trials
Full size table

Interventions of included trials

Thirty-nine trials (71%) consisted exclusively of single injection techniques. Of these, 27/39 compared different techniques or compositions of single injection ISB and 12/39 compared ISB with other blocks.20,21,22,23,24,25,26,27,28,29,30,31 Of the remaining 16 trials, 13 consisted solely of catheter techniques, including ten that compared ISB catheters of different technique or composition32,33,34,35,36,37,38,39,40,41 and three42,43,44 that compared ISB catheter with other catheter techniques. The remaining three trials compared ISB catheter to ISB single injection techniques45,46 and wound infiltration.47

A nerve stimulator was used as the sole technique for needle (and catheter) guidance for at least one study group in 16 trials.9,26,33,34,35,37,38,39,40,48,49,50,51,52,53,54 Three additional trials compared ISB with nerve stimulator to ultrasound guidance44,55 or paresthesia.56 Ultrasound approach was in plane, out of plane, or unclear in 29, five,24,31,42,43,57 and two45,58 of the remaining 36 trials, respectively.

Median and modal loading volume of block solution for catheter trials was 30 mL (range, 10–40 mL). Median and modal volume of single injection ISB was 20 mL (range 5–55 mL). Only one study compared different volumes of local anesthetic across ISB groups.59 Perineural epinephrine (dose range, 2.5–5 µg·mL-1) was used in 16 trials. Other perineural adjuvants administered per protocol included dexamethasone (eight trials; dose range, 1–8 mg),53,58,60,61,62,63,64,65 dexmedetomidine (two trials; dose range, 41.5–150 µg),57,66 sodium bicarbonate (two trials), magnesium sulphate (two trials),55,56,67,68 and triamcinolone acetonide (one trial).54 Additional detail on interventions is in ESM eAppendix.

Risk of bias

Only five trials were assessed as low risk of bias,28,29,63,66,69 14 trials were assessed as some concerns of bias,40,41,42,50,58,60,61,62,65,67,70,71,72,73 and 36 were judged as high risk of bias (Table 2).20,21,22,23,24,25,26,27,30,31,32,33,34,35,36,37,38,39,43,44,45,46,47,48,49,51,52,53,54,55,56,57,59,68,74,75 Only seven trials (13%) were assessed as low risk of bias for selective outcome reporting (domain 5). For the remaining 48 trials, there were some concerns of bias, typically due to an inability to confirm the reported PONS analysis was planned a priori. Some concerns of bias or high risk of bias due to deviations from the intended intervention were also common (domain 2, 44% of trials) and risk of bias due to measurement of the outcome (domain 4, 35% of trials). Both domains are influenced by insufficient reporting on blinding or lack of blinding of patients, caregivers, and outcome assessors.

Table 2 Risk of bias assessments for included studies
Full size table

Postoperative neurologic symptoms, hoarseness, and dyspnea outcomes

Only two out of 55 (4%) included trials were powered to determine the incidence of PONS as the primary outcome.55,56 Postoperative neurologic symptoms were assessed either by telephone interview (26/55, 47%), physical exam (21/55, 38%), or unclear methods (8/55, 15%). Repeated PONS assessments occurred in 13 trials including four that assessed for PONS three or four times.

Postoperative neurologic symptoms were most frequently assessed at one or two weeks, with the latest assessment at one year. Reported incidences of PONS tended to be highest at earlier assessment points, decreasing markedly between two weeks and one month (Table 3). No cases of PONS were reported in 38 trials with sample sizes ranging from 30 to 336 participants, including 19 trials with assessments at one week. At least one case of PONS (incidence range, 0.8–32%) was reported in 17 trials with sample sizes ranging from 41 to 910 participants, including ten with repeated PONS assessments (ESM eTable 2). Of 27 trials comparing single injection ISB with ISB of different technique or composition, 24 trials measured PONS between one and three weeks. Incidence ranged from 0% (19 trials, 30–256 participants per trial) to 32% (13/41 participants). Of 12 trials comparing single injection ISB with other regional techniques, ten reported PONS between one and four weeks. The incidence ranged from 0% (nine trials, 44–336 participants per trial) to 6% (4/60 participants). Of 16 trials comparing ISB catheter with other techniques or alternative ISB catheter technique or composition, ten reported PONS between one and three weeks. The incidence ranged from 0% (ten trials, 36–303 participants per trial) to 3% (2/62 participants).

Table 3 Incidence of PONS by postoperative follow-up time
Full size table

When summarized by type of surgery, in 23 trials of arthroscopy patients, PONS was measured between one and three weeks after surgery in all but three trials. No cases were reported in 15 trials (trial size range, 44–336 participants) and the incidence ranged from 1/58 (2%) to 13/41 (32%) in the remaining eight trials. In two trials of shoulder arthroplasty, PONS incidence ranged from 0/129 at three months to 2/70 (3%) at six weeks. No cases of PONS were reported among 302 patients undergoing open (nonarthroplasty) shoulder procedures across five trials assessing PONS between one week and one month after surgery. In the remaining 25 trials, the surgical populations were mixed or unclear. No PONS cases were reported in 17 (trial size range, 30–218 participants), of which 16 measured PONS between one and four weeks. In the remaining eight trials, the incidence ranged from 0.8% (2/240) at three months to 15% (15/99) at one week.

There were inconsistencies in how PONS diagnostic criteria were defined in the included trials (Table 4 and ESM eTable 2), ranging from ambiguous, nonspecific definitions (e.g., neurologic complications) to specific definitions (e.g., numbness, pain, tingling, or motor weakness in the operative arm). The most common definition criteria for PONS were the presence of one or more of paresthesia, sensory deficit, or motor deficit, used in 16/55 (29%) trials. Paresthesia (reported in 55% of trials), motor deficits (49% of trials), and numbness (40% of trials) were the most common criteria. Pain was a part of the diagnostic criteria in 18% of trials, and dysesthesia in 7% of trials. Of the included trials, 25% had nonspecific definitions of PONS and 9% were not classified (e.g., “other”).

Table 4 Frequency of specific neurologic symptoms used in PONS definitions for the included studies
Full size table

Of eight trials measuring dyspnea, four with a sample size range of 60–910 had zero outcomes.43,45,54,58 Two trials of ISB adjuncts by the same group reported dyspnea at two weeks in 7/280 (2.5%) patients63 and 2/198 (1.0%) patients,69 with neither study reporting persistent symptoms at six months. Another two studies reported dyspnea at three months in 1/69 (1.4%) patients47 and seven days in 2/120 (1.7%) patients.42

Of five trials measuring hoarseness, two trials54,58 of 218 and 910 patients each reported zero outcomes. The remaining three trials reported an incidence of hoarseness between one and four weeks of 1/120 (0.8%),42 7/280 (2.5%),63 and 17/198 (8.6%) patients.69 In the latter two trials, no cases of hoarseness persisted at six months.

Discussion

Statement of findings

We found that PONS were infrequently reported in RCTs of ISB use. Of 477 otherwise eligible trials, 422 (88%) were excluded solely for not measuring PONS. Further, across the 55 included studies that reported PONS, outcome definitions were inconsistent and results were usually at high risk of bias. Only 16/55 (29%) trials used the most common PONS definition. Consistent with concerns of bias and heterogeneity, incidence varied greatly across trials, even when comparing trials with similar follow-up periods. Because of these limitations in the existing literature, the predefined objectives related to quantifying the risk of PONS with ISB, including comparing ISB techniques and compositions, could not be achieved.

Contextualized with previous research

Randomized controlled trials are the only study design capable of balancing surgical, patient, and block related predictors of PONS across study groups, thereby providing the most unbiased estimates of PONS incidence in ISB vs in comparator blocks. Nevertheless, large observational studies4,5,11,12 comparing the incidence of PONS in ISB with other commonly used peripheral nerve blocks remain prominent in the literature despite their limitations. Specific details of block technique and composition are typically lacking, and comparisons of the incidence of PONS with ISB vs with other techniques are inadequately adjusted for confounding associations between type of block and other potential predictors of PONS. Additional smaller series of a few hundred patients receiving ISB76,77 provide estimates of PONS incidence but no comparison group. Reported incidence ranges in these studies vary but fall within the wide ranges of incidence reported in the randomized trials included in this review. This variation may be caused by variable definitions of PONS across studies, as was also found in the trials in this review. Where measured longitudinally, the incidence of PONS was typically highest soon after surgery and decreased over time, as was seen empirically across the trials in this review. The type of surgery and the surgical position may also influence the rate of PONS, as specific neurologic injuries are associated with different types of shoulder surgery.2

A recent systematic review of randomized trials comparing the analgesic effects of perineural dexamethasone vs placebo for any brachial plexus blockade also included “neurologic complications” as a secondary outcome. Just 8/33 trials reported neurologic complications, and events occurred in only one of these,15 echoing the challenges encountered in this review.

It could be argued that it is unreasonable to expect that an adverse outcome like PONS would be routinely measured with the same rigor as typical primary outcomes like analgesia. Nevertheless, we believe that, since ISB is a commonly used regional technique and neurologic complications are an important patient safety outcome,78 PONS outcome reporting should be improved and standardized. The development of a standardized PONS outcome would require input from and consensus among content experts, patients, and other stakeholders. The ideal standardized PONS outcome would be easily and reliably measured to facilitate consistent use in clinical research. The most commonly used and specific PONS definitions identified in this review would deserve scrutiny in this regard.

Strengths and limitations

The methods of this review were registered a priori and guided by the procedures of the Cochrane and PRISMA statements. The methodological and clinical heterogeneity of the existing literature was underestimated and prevented a meta-analysis that would quantify the risk of PONS with ISB vs other techniques, or among different ISB techniques. Variation in PONS definitions and follow-up time were the main sources of methodological heterogeneity while variation in the type of surgery, comparator groups, ISB technique, and composition contributed to clinical heterogeneity. Additionally, high risk of bias among included trials, possibly due to PONS being reported as a secondary outcome, would further limit the quality of evidence derived from a quantitative analysis. Inclusion of trials published in languages other than English may have increased the number of included trials but would likely not provide a sufficiently homogeneous pool of trials upon which to perform a meta-analysis.

Implications

This systematic review provides insight into the methodological deficiencies of the literature with regards to the measurement of PONS as an outcome, despite heterogeneity and high risk of bias preventing the originally planned quantitative assessment of the risk of PONS after ISB vs other techniques, or among ISB techniques. The development and implementation of a standardized PONS definition would help to improve the value of the individual trials and facilitate comparisons across trials, as was attempted in the current work and supported elsewhere.78 Additionally, routinely including PONS as an outcome in regional anesthesia trials, defining it in registries and blinding PONS outcome assessors to group assignment, would minimize bias in PONS outcome reporting. Together, these methodologic improvements could improve patient safety by identifying how surgical approach, ISB technique, and composition influence PONS outcomes.

Conclusion

Limitations in the existing literature prevented a quantitative comparison of the risk of PONS associated with ISB. This finding highlights the need for a standardized PONS outcome definition and follow-up time and improved consistency and methodologic rigor in how PONS outcomes are assessed.