Predicting Postsurgical Complications of Peripheral Nerve Decompression: NSQIP Analysis of Frailty Measures Versus Historic Proxies

Peripheral nerve decompression (PND) is a common, low-risk^1,2 soft tissue surgery typically performed at anatomic compression points in the hand and wrist. The most common nerve decompressions are performed to relieve pain or restore muscle function in the setting of focal median and ulnar nerve compression, such as in carpal tunnel syndrome or cubital tunnel syndrome. However, the general landscape of PND is changing to include less common sites and indications, including head and neck sensory nerve decompression for migraines.³ There are also Level 1 and 2 studies that strongly suggest that lower extremity nerve decompression in patients with diabetic neuropathy significantly decreases rates of ulcer formation, need for amputation, and associated health care costs.^4-6 Additionally, with newer regional and local anesthesia techniques, nerve decompression procedures have shifted away from hospitals and toward ambulatory surgery centers and even procedure rooms in an outpatient clinic setting.

Nerve decompression functional outcome studies are well-represented in the literature. However, few studies focus on short-term complications and their risk factors. Most studies that investigate complication risk factors examine hand surgery as a whole or carpal tunnel release in otherwise healthy patients,² but this may not necessarily accurately reflect the potentially higher risk for complications in patients undergoing the procedure at uncommon anatomic sites and for atypical indications. For example, there is a scarcity of data focused on lower extremity nerve decompression in patients with diabetic neuropathy and vasculopathy.

Given the changing landscape of nerve decompression, with frail patients being referred to ambulatory surgery centers to undergo procedures at uncommon anatomic sites, the ability to accurately predict postsurgical complications is valuable. Previously described risk factors, such as age, smoking status, and comorbidities, provide some insight, but current literature now explores frailty measures, such as the modified 5-Item Frailty Index (mFI-5) and modified Charlson Comorbidity Index (mCCI), as potentially stronger predictors of postsurgical complications in plastic and orthopedic surgery. This study sought to determine whether these frailty measures serve as better predictors of postsurgical complications in PND at less common anatomic sites when compared with previous risk proxies.

The authors performed a retrospective review of data from all patients in the American College of Surgeons’ National Surgical Quality Improvement Program (NSQIP) database who underwent PND from 2014 to 2019. Patients aged 18 or older were included if they underwent nerve decompression at the digits or extremities, as defined by Current Procedural Terminology (CPT) codes 64702, 64704, 64708, and 64722. Patients who underwent carpal tunnel release, cubital tunnel release, or Guyon’s canal release were excluded. Also excluded were those patients who underwent surgery of the brachial plexus or sciatic nerve proper, surgical treatment of neuromas or nerve lesions, surgery on sites other than the extremities, surgical repair of transected nerves, nerve grafting, or any other concurrent procedure.

Patient factors, such as age, body mass index (BMI), list of major comorbidities, and American Society of Anesthesiologists Physical Status Classification (ASA Class), were gathered and stratified. Age was classified as “young” (aged 18-49 years), “middle age” (aged 50-65 years), or “elderly” (aged > 65 years). BMI was classified as “non-overweight” (< 25.0 kg/m²), “overweight” (25.0-29.9 kg/m²), and “obese” (≥30.0 kg/m²). The mFI-5 score was calculated using the definition per Subramaniam et al.⁷ Many Charlson Comorbidity Index variants have been created; the authors opted to start with the original criteria by Charlson et al,⁸ then removed criteria for which NSQIP does not collect data (Table 1).

Data were gathered regarding postoperative 30-day complications. Each complication was assigned its Clavien-Dindo grade, and to produce a proxy of overall complication severity, each patient was given an aggregate Clavien-Dindo score (ACDS) by totaling the Clavien-Dindo grade for all complication occurrences (Table 2). Age, BMI, number of major comorbidities, ASA Class, mFI-5 score, and mCCI score were compared as predictors of all-cause 30-day complications, 30-day surgical site complications of any kind, length of stay, and complication severity (ACDS). Univariate and multivariate logistic regression was performed using the R statistical suite, version 4.1.0. Statistical significance was set at P < .05.

A total of 1120 patients were identified who underwent a PND meeting the study criteria between 2014 and 2019. The mean patient age was 51.3 (15.4) years. Patients had an approximately even sex distribution but were predominantly white (n = 728; 65.0%) and non-Hispanic (n = 798; 71.3%). Patients predominantly underwent neuroplasty of major arm or leg nerves, CPT 64708 (n = 506; 45.2%), or decompression of other nerves, CPT 64722 (n = 436; 38.9%) (Table 3). Patients were predominantly healthy, with most having no comorbidities (n = 613; 54.7%) or 1 major comorbidity (n = 328; 29.3%). Patients were also generally not frail, with over half of patients having an mFI-5 score of 0 (n = 636; 56.8%) or an mCCI score of 0 (n = 373; 33.3%) or 1 (n = 208; 16.8%). The predominant comorbidities were medicated hypertension (n = 426; 38.0%), diabetes (n = 196; 17.5%), and chronic steroid use (n =33; 2.9%). One-fifth of patients were smokers (n = 239; 21.3%), had a mean BMI of 30.6 (7.0) kg/m², and a median BMI of 29.7 kg/m² (Table 4).

Complications were rare, with a total complication rate of 3.4% (38 patients, 43 complications). In 1120 patients, the most common complication was superficial surgical site infection (n = 17; 1.5%). There were 11 readmissions (1.0%) and 10 unplanned returns to the operating room (0.9%). The mean ACDS for patients who suffered complications was 3.4 (2.6). ACDS included any surgical site complication, systemic complication, reoperation, and readmission as recorded in the ACS-NSQIP database. While nearly all patients were discharged the same day, 71 patients (6.3%) remained overnight and 60 patients (5.4%) were admitted for 2 or more days. The median length of stay for non-outpatient surgery was 1.0 (2.0) days (Table 5). When the mFI-5 score was analyzed within the context of postoperative complications in univariate analysis, increasing mFI-5 score was significantly associated with readmission (P = .014) and all-cause complications (P = .012), and this association approached significance for surgical site complications (P = .063) (Table 6). 

On univariate analysis, several risk factors and indices were predictive of complications. The strongest predictor of all-cause 30-day complications was ≥ 3 major comorbidities (odds ratio [OR], 95% CI: 6.26, 1.36-21.32; P = .007), while mFI-5 score, mCCI score, and fewer major comorbidities were also predictive of all-cause complications. Complication severity was associated with number of major comorbidities and mFI-5 score, while length of stay was most strongly predicted by age ≥ 65 years (OR, 95% CI: 2.17, 1.37-3.42; P = .0008) and mCCI score of 3 (OR, 95% CI: 1.77, 1.01-3.05; P = .041). The only factor predictive of surgical site complications was ≥ 3 major comorbidities (OR, 95% CI: 6.94, 1.00-30.44; P = .019); age, diabetes, obesity, smoking status, and frailty were not associated with an increased risk. The only risk factor associated with reoperation rates was smoking status (OR, 95% CI: 3.74, 1.03-13.56; P = 0.038), and the only risk factor associated with readmission rates was an mFI-5 score of 1 (OR, 95% CI: 7.00, 1.68-47.16; P = 0.016) (Table 7).

In multivariate analysis, mFI-5 score (P = .022) and mCCI score (P = .038) were the only predictors of any complication; age, BMI, smoking status, diabetes status, ASA class, and steroid use were not significantly associated with postoperative complications.

Current Literature

There are sparse data regarding postsurgical non-nerve outcomes and complications after PND, particularly outside of the distal arm and hand. As such, the present literature review was supplemented with soft tissue hand surgery and PND of the distal arm and hand as proxies for general decompression of superficial peripheral nerves. These studies are only partially applicable, especially given that the present study seeks to examine atypical nerve decompressions in patients who are potentially substantially less healthy than their counterparts undergoing PND in the hand. However, given the limits of the current state of the literature, the authors feel this proxy relationship and the current study are a reasonable first step.

Some studies relating to common PNDs can contextualize risks associated with PNDs at less common sites and in less healthy patients. A study of approximately 850,000 carpal tunnel release patients in the United Kingdom using National Health Service (NHS) data showed an all-cause 90-day complication rate of 0.082%, with even lower rates of dehiscence and infection.¹ This extremely low complication profile may be partially generalizable to PNDs with limited incisions, and it suggests that the number of patients needed to identify true risk factors would be prohibitively high. The use of a procedure room,⁹ WALANT (wide awake, local anesthesia, no tourniquet) techniques,¹⁰ or foregoing postoperative antibiotics¹¹ do not increase risk. Some risk factors still remain uncertain—diabetics may have worse recovery,¹² similar recovery,¹³ or a mixed recovery,¹⁴ although hemoglobin A1c values do appear to accurately reflect risk.¹⁵ On the other hand, some risk factors are extremely clear: steroid injection within 30 days of nerve decompression increases risk of infection, wound complications, and 1-year reoperation needs.^16,17

 Studies that examine other PNDs, or all nerve surgery in aggregate, may not be completely applicable because some of these studies include traumatic injury repair, nerve grafting, or brachial plexus and lumbosacral plexus/sciatic nerve surgery. However, these studies suggest an increased risk with prolonged surgery, high ASA status, bleeding disorders, and higher BMI.^18,19 Studies of isolated cubital tunnel release suggest increasing risk with male sex, obesity, smoking status, and such comorbidities as lung and kidney disease.²⁰

Contextualized Results

At first glance, PND in diabetic patients with peripheral vascular disease and diabetic neuropathy may seem to be a much higher-risk proposition than nerve decompression in the hand. However, the opposite case appears to be true. There is strong Level 1 and 2 evidence²¹ from prospective and randomized controlled trials that suggests that PND in the lower extremity improves protective sensation²² and pain^23,24; improves wound healing; decreases the risk of diabetic foot wounds⁴ or accelerates healing of existing diabetic foot ulcers^2,25 in patients with prior wounds²⁶; and significantly reduces hospitalization length of stay, duration of wound care, and the need for amputation.^6,27 Compared with medical management, preliminary simulation studies of PND in diabetic patients suggest the procedure may yield higher quality-adjusted life-years, reduced need for amputation, improved survival, and higher net monetary benefits and cost savings.⁵ 

The present study showed a complication rate of 3.4%, which is substantially higher than the NHS-reported 0.082% in carpal tunnel surgery. However, the NSQIP database reflects a generally more ill and more complex patient distribution. These patients undergo surgery in the hospital setting; typically, patients who require hospital services for PND are those who are considered too high-risk for procedure rooms or ambulatory surgery centers. While the strongest predictor of all-cause complications was 3 or more major comorbidities, mFI-5 and mCCI were also predictive of multiple outcomes and approached significance for multiple additional outcomes that did not otherwise have strong predictors. An appropriately powered study could potentially identify if these are statistically significant.

This study has several key limitations. While the use of NSQIP data is widely accepted in plastic surgery, orthopedic surgery, and hand surgery, the NSQIP database is designed for much more invasive procedures and does not measure many outcomes that would be relevant to nerve decompression.¹⁰ Patient selection bias is strong, as NSQIP excludes WALANT procedures performed outside of an operating room, patients seen in a private practice, and low-risk patients who do not undergo surgery at participating hospitals. The data within NSQIP does not provide adequate granularity; for example, NSQIP does not stratify diabetes control by hemoglobin A1c or by insulin needs,²¹ nor does it quantify the amount and recency of smoking.²⁸ Outside of the limits of NSQIP, surgeons may have different policies regarding canceling surgery for patients who have recently smoked, and some surgeons may have cutoff points for BMI or hemoglobin A1c above which they do not offer surgery to patients. The Clavien-Dindo complication grade, in particular, is designed to allow cross-comparison of complications, not to create a total score of aggregate complication severity; furthermore, the literature does not describe ACDS as reported in this paper.

Frailty measures, such as the mFI-5 and mCCI, are able to predict increased risk of 30-day all-cause complications, complication severity, length of stay, and readmission. However, the risk profile of PND outside of the distal arm and wrist remain poorly described in the literature. Further work is needed to clarify specific frailty components and to adequately power a determination of other related risk factors.