Specialty Mediated 30-Day Complications in First Rib Resection for Thoracic Outlet Syndrome

Abstract

Background:

Thoracic outlet syndrome (TOS) takes on heterogenous upper extremity manifestations depending on whether the artery, vein or brachial plexus is primarily compressed. As a result of these variable vascular and neurogenic symptoms, these patients present to surgeons of various training backgrounds for surgical decompression. Surgeon specialty is known to correlate with outcomes for numerous vascular procedures, but its role in TOS is unclear. In this work we examine the association of surgeon specialty with short-term outcomes following first rib resection (FRRS) for TOS.

Methods:

Using the American College of Surgeons National Surgical Quality Improvement Program database, 3,070 patients were identified who underwent FRRS for TOS between 2006–2017. The primary outcomes of the study were 30-d complications, including postoperative hemorrhage requiring transfusion, wound complications, pneumothorax and deep venous thrombosis. Arterial, venous, and neurogenic TOS were distinguished with ICD-9 and 10 codes while patient characteristics, provider specialty, and postoperative outcomes were classified through a combination of standard National Surgical Quality Improvement Program variables and ICD data.

Results:

Most FRRS were performed by vascular surgeons (87.9%), general (6.9%) and thoracic surgeons (4.4%). The relative distribution of vascular TOS between the specialties was not significantly different, with non-vascular surgeons performing an equivalent amount of FRRS for arterial (1.1% versus 2.4%) and venous TOS (8.6% versus 9.1%, both P > 0.05). Patients who underwent FRRS with non-vascular surgeons experienced more frequent perioperative transfusions (3.2% versus 1.2%, P = 0.001) and wound infections (1.9% versus 0.8%, P = 0.04). On multivariable regression, patients undergoing FRRS for venous TOS were more likely to require blood transfusion (odds ratios:3.63, 95% CI 1.43–9.25). Patients operated on by surgeons whose specialty was not among the top three most common specialties performing FRRS had a 40% longer operative time (incidence rate ratios:1.42, 95% CI 1.15–1.74) as well as a significantly increased odds of requiring a transfusion (odds ratios:9.87, 95% CI 2.28–42.68).

Conclusions:

The significantly increased operative times and transfusion requirements associated with specialties who uncommonly perform FRRS suggest the role of surgeon experience and volume in this procedure may play more of a role than specialty training. These data also suggest that vascular TOS carries unique risks that should be kept in mind when performing FRRS.

Introduction

Thoracic outlet syndrome (TOS) can manifest as a heterogeneous set of clinical presentations due to the distinct major structures which pass through the thoracic outlet.^1,2 Individual compression of the subclavian artery, vein, or brachial plexus can trigger a unique set of symptoms that in turn may dictate the type of surgeon these patients initially present to surgical decompression.³ Decompression of the thoracic outlet can be achieved through a number of different surgical procedures, but most frequently entails first rib resection (FRRS) via a transaxillary or supraclavicular approach.^4–5 The majority of these procedures are performed by vascular surgeons; however, it is not uncommon for general, thoracic, orthopedic or neurosurgeons to perform these procedures as well.^5–6

Surgeon specialty and training background has been shown to correlate with 30-d outcomes for several surgical procedures associated with cross-disciplinary presentation.^7–10 For carotid endarterectomy, multiple studies have demonstrated that patients who are operated on by vascular surgeons have improved outcomes compared to general surgeons, but also that operator specialty dictates the comorbidity burden of the patient population.^9–10 Overall, data are indicative of improved outcomes when vascular surgeons perform vascular procedures in high risk or complex patients, which is thought to be indicative of a volume-outcomes relationship but also a familiarity with the underlying vascular anatomy in complex cases.⁸ However, not all surgical procedures demonstrate improved outcomes with the involvement of specialists^11–12 and the role of specialist referral versus prompt surgical management remains to be defined for many procedures.

Despite the fact that the skills required to perform thoracic outlet decompression should translate to all types of TOS, the various types present distinct intraoperative pitfalls and unique postoperative courses. Perioperative outcomes of FRRS are associated with the etiology of TOS, with vascular TOS patients having a higher rate of perioperative complications when compared with neurogenic TOS patients.^5–6 There are currently no published studies that report the distribution of TOS operated on and the outcomes of the various types of TOS in the hands of surgeons of different training backgrounds. Here we utilize a large national database to elucidate specialty-specific differences in the patient populations and short-term outcomes of FRRS for TOS.

Methods

Study design

This is a retrospective analysis utilizing data from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database for the time period 2006 to 2017. The ACS-NSQIP is a prospectively maintained database that incorporates data from >500 hospitals who participate voluntarily as part of a national quality improvement effort focused on identifying 30-d outcomes of elective surgical procedures. The sampling methodology of NSQIP is systematic and entries are reviewed on a cyclic basis in order to reduce bias in terms of case selection and case mix. The NSQIP database also utilizes certified clinical reviewers to enter data, reducing some inaccuracies associated with administrative databases. The Institutional Review Board of The Johns Hopkins University School of Medicine approved the study and waived informed consent requirements given its use of deidentified data.

The study cohort included adult patients (≥18 y of age) who underwent FRRS. The patient cohort was identified using the unique Current Procedural Terminology code for FRRS (21615), which includes cervical rib excisions. In order to identify the TOS subtypes, ICD-9 and ICD-10 diagnosis codes were applied to the identified FRRS cohort to classify cases as arterial (aTOS), venous (vTOS), or neurogenic (nTOS) (Supplemental Table 1). Cases of FRRS that did not fall into the any of these three categories were not excluded but were classified as undifferentiated TOS. The specialty of the operating surgeon is a standard recorded variable in NSQIP; within the FRRS cohort, procedures were performed by surgeons with a training background in vascular, general, thoracic, plastics, orthopedics, otolaryngology (ENT), cardiac and neurosurgery. These latter five specialties all performed <1.0% of total FRRS cases and were grouped together for subgroup analysis.

Outcomes

Primary outcomes of interest were 30-d postoperative complications, which included pneumothorax, hemothorax, bleeding requiring transfusion, deep venous thrombosis, peripheral nerve injury and return to the operating room. All of these endpoints were identified with a combination of NSQIP variables and ICD data and compared with respect to operator surgical specialty. Secondary outcomes included operative factors such as operating room time and estimated blood loss, as well as in-hospital length of stay.

Statistical analysis

We described the patient populations operated on by the different surgical specialties with descriptive statistics. Outcomes for each type of TOS were calculated using χ² and ANOVA testing for categorical and continuous variables, respectively. In order to determine the association of baseline patient and operator factors with the primary outcomes, multivariable logistic regression modeling was performed controlling for all factors listed in Table 1, including baseline comorbidities, demographics, and TOS type. Postoperative complications assessed with multivariable Results of multivariable logistic regression was described as odds ratios (OR) for categorical variables and incidence rate ratios (IRR) for continuous variables. All statistical analyses were performed using Stata Version 14.2 (StataCorp LP, College Station, TX). Values were recorded as statistically significant at a level of α ≤ 0.05.

Table 1 –

Demographic and comorbidities of patients undergoing first rib resection by operator specialty.

	Vascular n = 2,697 (87.9%)	General n = 211 (6.9%)	Thoracic n = 134 (4.4%)	Other n = 28 (0.9%)	P value
Mean age, y (IQR)	35.5 (25,44)	39.3 (29,48)	40.1 (28,49)	38.8 (26, 51)	<0.001
Female sex	1,597 (59.3%)	135 (64.0%)	68 (50.8%)	19 (67.9%)	0.078
Race					0.042
White	2,281 (87.0%)	186 (92.1%)	108 (80.6%)	23 (88.5%)
Black	92 (3.5%)	5 (2.5%)	3 (2.2%)	2 (7.7%)
Other non-white	249 (9.5%)	11 (5.5%)	23 (17.2%)	1 (3.9%)
TOS Type					<0.001
Venous	245 (9.1%)	21 (10.0%)	10 (7.5%)	1 (3.6%)
Arterial	65 (2.4%)	3 (1.4%)	1 (0.8%)	0 (0.0%)
Neurogenic	2,268 (84.1%)	159 (75.4%)	96 (71.6%)	19 (67.9%)
Undifferentiated	119 (4.4%)	28 (13.3%)	27 (20.2%)	8 (28.6%)
Comorbidities
Hypertension	344 (12.8%)	22 (10.4%)	23 (17.2%)	6 (21.4%)	0.158
Diabetes	86 (3.2%)	4 (1.9%)	7 (5.2%)	1 (3.6%)	0.399
Current smokers	494 (18.3%)	48 (22.8%)	34 (25.4%)	5 (17.9%)	0.097
History of COPD	29 (1.1%)	3 (1.4%)	2 (1.5%)	0 (0.0%)	0.871
ASA Class					<0.001
Class 1	497 (18.4%)	31 (14.7%)	17 (12.7%)	6 (21.4%)
Class 2	1,788 (66.3%)	146 (69.2%)	79 (59.0%)	17 (60.7%)
Class 3	378 (14.0%)	31 (14.7%)	34 (25.4%)	4 (14.3%)
Class 4	33 (1.2%)	3 (1.4%)	3 (2.2%)	1 (3.6%)
Class 5	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (0.8%)

ASA Class = American Society of Anesthesiologists Classification; COPD = chronic obstructive pulmonary disease; IQR = Interquartile Range.

Bolded values are statistically significant at a level of p less than or equal to 0.05.

Results

Overall, a total of 3,070 FRRS were identified in the NSQIP database; the demographic information of these patients is summarized in Table 1. Of these patients, the majority (87.9%) were performed by vascular surgeons, followed by general (6.9%) and thoracic surgeons (4.4%). The practice patterns between the specialties with respect to vascular TOS were not significantly different, with thoracic and general surgeons performing an equivalent amount of arterial and venous TOS (0%−1.4% versus 2.4% and 7.5%−10.0% versus 9.1%, respectively) when compared with vascular surgeons (both P > 0.05). Vascular surgeons operated on a larger proportion of neurogenic TOS (84.1%) when compared to general and thoracic surgeons (71.6%−75.4%). Thoracic surgeons operated on significantly more patients with an ASA class ≥III (27.8%) when compared to the other specialty groups (15.2%−20.7%).

An analysis of postoperative outcomes by specialty is demonstrated in Table 2 and Supplemental Table 2. Patients who underwent FRRS with non-vascular surgeons experienced more frequent wound infections (1.9% versus 0.8%, P = 0.04), a difference which was attributable to the general surgery group (2.8%, P = 0.023) on subgroup analysis. There was significantly higher use of postoperative transfusion among patients undergoing surgery by a non-vascular surgeon (3.2% versus 1.2%, P = 0.001), which was significantly higher than the thoracic surgery group (5.2%) specifically. There was also a significantly higher rate of pneumonia in the group of patients receiving FRRS by surgeons with a specialty performing <1% of (7.1% versus <1%, P < 0.001). There were no differences elicited with respect to reintubation, return to the operating room, pneumothorax, deep vein thrombosis, arterial thrombosis, or peripheral nerve injury between the various surgical specialties. The mean LOS for patients undergoing FRRS with a vascular surgeon was 2.7 d, which was shorter than both the thoracic and other composite group (3.3 and 3.5 d, respectively; P = 0.003 and 0.068).

Table 2 –

Postoperative complications and operative information following first rib resection by operator specialty.

	Vascular n = 2,697 (87.9%)	General n = 211 (6.9%)	Thoracic n = 134 (4.4%)	Other n = 28 (0.9%)	P value
Length of stay, d (95% CI)	2.69 (2.60–2.78)	2.61 (2.21–3.01)	3.26 (2.74–3.79)	3.46 (1.82–5.11)	0.028
Surgical site infection	21 (0.78%)	6 (2.84%)	1 (0.75%)	0	0.023
Pneumonia	16 (0.59%)	1 (0.47%)	0	2 (7.14%)	<0.001
Reintubation	6 (0.22%)	2 (0.95%)	0	0	0.221
Deep venous thrombosis	46 (1.71%)	2 (0.95%)	1 (0.75%)	0	0.606
Return to OR	148 (5.49%)	9 (4.27%)	10 (7.46%)	0	0.354
Postoperative transfusion	31 (1.15%)	2 (0.95%)	7 (5.22%)	3 (10.71%)	<0.001
Pneumothorax	3 (0.11%)	1 (0.47%)	0	0	0.532
Hemothorax	5 (0.19%)	0	0	0	0.875
Arterial thrombosis	1 (0.04%)	0	0	0	0.987
Peripheral nerve injury	2 (0.07%)	0	0	0	0.964
Operative time, min (95% CI)	149 (146–153)	146 (132–160)	161 (138–183)	211 (132–290)	0.004

95% CI = 95% confidence interval.

Results of mutivariable modeling are displayed in Table 3. On multivariable regression, patients receiving FRRS for vTOS were more likely to experience an episode of postoperative hemorrhage requiring transfusion (OR:3.63, 95% CI 1.43–9.25). Arterial TOS patients were significantly more likely to require a return to the OR (OR:4.43 95% CI 2.22–8.84) and had longer operative times (IRR:1.65 95% CI 1.44–1.87). Performance of FRRS by a general surgeon remained independently associated with surgical site infection (OR:3.31, 95% CI 1.17–9.33). Patients operated on by surgeons whose specialty was not among the top three most common specialties performing FRRS had a 24% longer operative time (IRR:1.24, 95% CI 1.01–1.54) and a significantly increased risk of postoperative hemorrhage (OR:9.87, 95% CI 2.28–42.68).

Table 3 –

Multivariable predictors of select postoperative complications.

	LOS, d (IRR)	Operative time, min (IRR)	Postoperative transfusion (OR)	Surgical site infection (OR)	Return to operating room (OR)
Age	1.00 (1.00–1.00)	1.00 (1.00–1.00)	1.03 (1.00–1.05)	1.00 (0.97–1.04)	1.00 (0.91–1.02)
Sex	0.95 (0.90–1.00)	0.85 (0.82–0.88)	0.56 (0.28–1.11)	0.78 (0.36–1.77)	0.56 (0.40–0.78)
Race
Black	0.98 (0.84–1.14)	1.06 (0.96–1.19)	2.08 (0.60–7.20)	-	0.96 (0.41–2.26)
Other non-white	1.25 (1.00–1.56)	1.02 (0.86–1.21)	3.99 (0.73–21.76)	2.89 (0.36–23.25)	2.42 (0.89–6.61)
Surgical specialty
General	0.92 (0.82–1.03)	0.93 (0.86–1.00)	0.51 (0.10–2.44)	3.31 (1.17–9.32)	0.84 (0.42–1.71)
Thoracic	1.06 (0.93–1.20)	0.99 (0.90–1.09)	2.07 (0.75–5.75)	0.83 (0.10–6.58)	1.28 (0.64–2.58)
Other	1.06 (0.80–1.40)	1.24 (1.01–1.54)	9.87 (2.28–42.68)	-	-
TOS Type
Venous	1.13 (1.02–1.24)	1.04 (0.97–1.11)	3.63 (1.43–9.25)	0.87 (0.20–3.84)	1.43 (0.86–2.37)
Arterial	1.49 (1.27–1.76)	1.64 (1.44–1.87)	3.59 (0.93–13.80)	1.47 (0.18–11.96)	4.45 (2.23–8.88)
Undifferentiated	1.42 (1.27–1.59)	1.36 (1.24–1.48)	3.60 (1.49–8.68)	1.19 (0.28–5.06)	1.47 (0.79–2.74)
Comorbidities
Hypertension	1.09 (1.00–1.19)	1.06 (0.99–1.13)	0.61 (0.24–1.53)	0.53 (0.14–1.99)	1.13 (0.68–1.90)
Current Smoker	0.96 (0.89–1.03)	0.95 (0.90–0.99)	1.12 (0.50–2.49)	1.01 (0.38–2.67)	0.57 (0.35–0.93)
Diabetes
Insulin dependent	1.06 (0.87–1.31)	0.84 (0.71–0.99)	0.85 (0.19–3.82)	1.33 (0.12–14.89)	0.44 (0.11–1.73)
Non-insulin dependent	1.02 (0.83–1.25)	0.99 (0.84–1.15)	0.39 (0.04–3.77)	6.03 (1.15–31.40)	2.07 (0.83–5.15)
COPD	1.15 (0.90–1.45)	1.03 (0.85–1.24)	0.55 (0.06–5.11)	4.14 (0.76–22.48)	1.36 (0.36–5.08)
ASA Class
Class 1	1.01 (0.94–1.09)	1.01 (0.96–1.07)	0.91 (0.25–3.26)	1.18 (0.37–3.78)	1.48 (0.97–2.27)
Class 3	1.20 (1.11–1.30)	1.04 (0.98–1.11)	3.06 (1.33–6.99)	2.08 (0.72–5.95)	1.32 (0.82–2.14)
Class 4	2.22 (1.81–2.73)	0.93 (0.77–1.11)	13.78 (4.09–46.54)	6.70 (0.97–46.25)	6.14 (2.42–15.51)
Class 5	1.19 (0.36–3.98)	1.89 (0.66–5.47)	-	-	-

Discussion

A specialty-outcomes relationship has been previously demonstrated for a number of surgical procedures which have cross-disciplinary management. In the present study, we sought to in define the association of specialty with short-term outcomes of FRRS for TOS, a phenomenon which has a unique range of manifestations and, subsequently, a wide range of surgical specialists who care for these patients. We found there was an equivalent level of vascular complexity in the case mix of non-vascular surgeons and that there was a significantly higher rate of specific complications when patients were operated on by non-vascular surgeons, including postoperative hemorrhage, to which venous TOS patients were uniquely susceptible. Overall, the data here suggest FRRS for TOS has fewer severe hemorrhagic complications requiring blood transfusion in the hands of vascular surgeons. These data also reaffirm prior research suggesting that despite requiring a similar surgical approach, vascular TOS is a surgical entity distinct from neurogenic TOS and likely benefits from the involvement of a vascular surgeon.

A morbidity and mortality benefit associated with the performance of operations by vascular surgeons has been reported for multiple surgical procedures with potentially complex vascular anatomy, including below-knee amputation,⁷ sartorius muscle flaps⁸ and, most prominently, carotid endarterectomy.^9–10,13 In carotid endarterectomy, the specialty-specific stroke-related morbidity benefit of vascular surgeon involvement has recently been shown to be conferred primarily on asymptomatic carotid stenosis patients rather than on symptomatic patients.¹³ This would suggest the contribution of the volume-outcome relationship which has been well established for many surgical procedures; that is, as individual surgeon procedural volume increases, that surgeon’s outcomes improve accordingly.^14–15 In the interrogation of the present dataset, a similar phenomenon is noted. When examining unadjusted associations between specialty and postoperative outcomes, operation by a non-vascular surgeon was shown to be associated with a significantly higher rate of postoperative transfusion. After adjusting for confounders, the increased risk of hemorrhage and prolonged operative time remained significantly increased in the group of specialties who perform FRRS with the least frequency. These data suggest at the very least that TOS in general is best handled in the hands of those surgeons who care for it most frequently.

It is well-established that the most common manifestation of TOS is neurogenic and the majority of FRRS is performed for a neurogenic indication.^5,16 This is reflected in the present interrogation of the NSQIP dataset, in which 82.8% of patients received FRRS for nTOS. It has previously been demonstrated that the perioperative outcomes of FRRS are associated with the etiology of TOS. Despite undergoing a similar decompression, patients undergoing surgical decompression for vascular TOS are known to have longer lengths of stay and operative length as well as a higher likelihood of needing to return to the OR when compared to patients undergoing the operation for neurogenic TOS.^5–6 This is logical, as it is known that distinct anatomic abnormalities correspond with the different subtypes of TOS.^17–18

TOS with overt venous or arterial manifestations presents a greater challenge to the surgeon for a number of reasons.¹⁹ Arterial TOS frequently presents with a post stenotic aneurysmal change of the subclavian artery with partial or complete thrombosis, requiring preoperative anticoagulation and operative subclavian replacement to address the aneurysmal degeneration of the artery. The urgent nature of these operations as well as the addition of a bypass grafting procedure grants them greater surgical complexity and higher operative risk.^20–21 The operative risks associated with venous TOS are more subtle and largely stem from the need for preoperative anticoagulation to address subclavian venous thrombosis and the need to restart that anticoagulation postoperatively.²² We see in this dataset a similar higher level of risk conferred on the vascular TOS patients, with aTOS patients having a longer operative time and higher risk for return to the OR (and vTOS patients with a significantly higher rate of postoperative hemorrhage, which is likely related to perioperative management of anticoagulation. While the complication rates shown here do not indicate that postoperative bleeding frequently requires a return to the OR, perioperative blood transfusion has shown to be associated with multiple risks and should be avoided where possible.^23–24

This work should serve to inform surgeons of potential pitfalls of operating on the various types of TOS and generate information about patient profiles that may be more appropriate for specific surgical specialty referral. It seems clear that the bulk of aTOS patients should be cared for by vascular surgeons when possible given the high likelihood of need for bypass grafting ^17,20; likewise, vTOS patients with a high risk of perioperative hemorrhage, need for systematic restarting of systemic anticoagulation and subsequent potential need for venography and venoplasty would also likely benefit from vascular surgery involvement.²⁵ In nTOS, this is less clear; the surgical treatment of TOS has historically been somewhat controversial, particularly with regard to neurogenic TOS.¹⁶ While much has been made here and in the literature about specific classifications of TOS, it is important to acknowledge the intimate association of the various structures within the thoracic outlet and that they are rarely compressed individually.²¹ While patients may not present to a healthcare provider complaining of symptoms stemming from a vascular etiology of TOS, this is not a guarantee that the vascular structures are uninvolved. A retrospective study of patients treated at a single center for nTOS demonstrated a 5.2% rate of incidental venous or arterial involvement on further testing.²⁶ Further investigation into the long-term functional outcomes of these patients is needed before specialty-specific recommendations can be made regarding the need for referral of these patients.

There are a number of limitations in this study, many of which are inherent to the use of the NSQIP dataset itself, including the lack of data on surgeon-specific case volume, which may influence outcomes more than provider specialty alone, as well as the approach to FRRS. It has been previously postulated that FRRS is under-represented in NSQIP⁶; indeed, there are a limited number of venous and arterial TOS cases represented here, which limits the power of this study to perform specialty-specific subgroup analysis for each type of TOS. Likewise, there are no unique Current Procedural Terminology codes or ICD diagnosis codes for arterial, venous, and neurogenic TOS; these categories were generated within the FRRS cohort with the use of surrogate ICD diagnosis codes, which introduces the potential for misdiagnosis. Additionally, while the focus of this work is primarily on short-term complications, granular long term functional outcomes and regional availability of TOS specialists may be a key focus for future efforts seeking to define appropriate referral patterns for TOS, particularly nTOS. A major strength of using this dataset is that it captures both inpatient and outpatient surgical procedures, a point particularly germane to FRRS. FRRS is often performed under outpatient status with an extended overnight in-hospital recovery period of <24 h, causing many of these procedures to be lost in databases which purely sample inpatients.

Conclusions

The significantly increased operative times and transfusion requirements associated with specialties who uncommonly perform FRRS suggest the role of surgeon experience and volume in this procedure may play more of a role than specialty training. However, these data also suggest that vascular TOS carries unique hemorrhagic risks that should be kept in mind when performing FRRS; despite that, vascular surgeons did not perform a statistically higher proportion of these cases than other specialties. Referral should be considered for cases of TOS with known vascular involvement. Additional investigations into the long-term functional outcomes of FRRS for nTOS are needed in order to determine the need for referral of this particular subgroup.