Orthopaedics and neurosurgery: Is there a difference in surgical outcomes following anterior cervical spinal fusion?

Abstract

Objective

The superiority of neurosurgical over orthopaedic spinal procedures is a point of contention. While there is the perception that neurosurgeons are more specifically trained to deal with spinal pathology, no study has directly compared outcomes of spinal surgeries performed by both groups.

Methods

We sought to evaluate the differences in length of surgery, hospital stay, complications, mortality, and readmission for anterior cervical decompression and fusion (ACDF) performed by neurosurgeons versus orthopaedic surgeons.

Results

17,967 ACDF procedures were analyzed. Neurosurgeons performed 74.3% of the fusions with a trend towards longer operative times and significantly more patients that were discharged to extended care facilities. There was no significant difference in the length of stay, overall complications, mortality, readmission, or reoperation when comparing the two specialties.

Conclusion

Despite a significantly higher volume of ACDF performed by neurosurgeons, outcomes are comparable following orthopaedic and neurosurgical procedures.

1. Introduction

The rates of elective spinal procedures for degenerative cervical spondylosis have steadily risen over the past forty years.^1,2 While there has been a general increase in patient age and comorbidity burden, advances in technology and practice patterns have provided for reduced hospitalization times and exceedingly low complication rates.2, 3, 4, 5, 6, 7

Previous analyses support a steady increase in procedural volume, with minimal difference in postoperative outcomes between orthopaedic and neurosurgical specialties despite this evolving patient demographic.8, 9, 10, 11, 12 However, a more contemporary analysis is necessary given the rapidly changing training environment, as well as the dynamic nature of imaging technology, surgical techniques, healthcare delivery, and patient expectations.^9,13

Despite ongoing improvements in the quality of surgical intervention, differences in training and practice patterns amongst spine surgeons may ultimately have an influence on outcomes.13, 14, 15 The current study sought to evaluate the differences in specialty-related volume, patient-centered surgical outcomes, and complications following ACDF performed by orthopaedic versus neurosurgeons.

2. Materials and methods

The National Sample Program (NSP) of the American College of Surgeons (ACS) National Surgical Quality Improvement Program (ACS-NSQIP) was queried to identify all patients who underwent isolated ACDF [Current Procedure Terminology (CPT) Code 22551] between 2011 and 2014. The NSQIP is a prospective, multi-institutional database that represents a systems-based approach to surgical quality improvement within the United States, and can be utilized to measure risk-adjusted surgical outcomes.16, 17, 18, 19, 20, 21 Patients were included if they were at least 18 years or older, if the primary surgical specialty of the attending surgeon was orthopaedics or neurosurgery, and if outcome data was recorded.

Demographic information (age, sex, and body mass index), comorbidities (diabetes, cardiac disease, American Society of Anesthesiologists (ASA) classification), operative characteristics, and outcomes were collected and directly compared between the two-subspecialty groups. The primary outcomes of interest were hospital and ICU length of stay, 30-day post-operative complications, mortality, unplanned reoperation, and readmission. Postoperative complications were categorized as overall, systemic, and local complications. Mortality was recorded separately. Overall complications were defined as incurring any systemic and/or local complications. Systemic and local complications were further stratified into major and minor. Major systemic complications were defined as complications related to medical conditions that developed from nonsurgical causes and included: (1) thromboembolic (pulmonary embolism); (2) pulmonary (unplanned intubation, ventilator-assisted respiration >48 h); (3) sepsis (postoperative sepsis or septic shock); (4) cardiovascular (myocardial infarction, cardiac arrest requiring CPR, cerebrovascular accident or stroke); (5) renal (acute renal failure); and (6) coma. Minor systemic complications were defined as complications related to medical conditions that developed from nonsurgical causes and included: (1) urinary tract infection; (2) thromboembolic (deep venous thrombosis); (3) pulmonary (pneumonia); and (4) renal (progressive renal insufficiency). Major local complications included surgical site wound infection by anatomical location (deep, and/or organ/space SSI), peripheral nerve injury, and graft/prosthesis failure. Minor local complications included surgical site wound infection by anatomical location (superficial) and wound disruption. Reoperation was defined as a return to the operating room within 30 days of the index operation for intervention of any kind.

2.1. Statistical analysis

The statistical means and standard deviations were calculated for continuous variables. Categorical data was expressed as frequencies. The Chi-square test was used to characterize categorical variables. Univariate analysis was utilized to evaluate differences between outcomes of orthopaedic and neurosurgical cases. In all cases, a P-value of 0.05 or less was considered statistically significant. Propensity score matching was performed to minimize the potential risk of bias inherent to the nonrandomized design of this study, allowing for an improved estimate of treatment effect by simultaneously balancing observed covariates between the 2 cohorts. Matching was completed using a 1:1 nearest neighbor matching algorithm with computerized greedy matching without replacement. For each randomly selected case from the orthopaedic cohort, one case from the neurosurgical cohort with the closest propensity score was matched for preoperative characteristics. After a match is made, the corresponding Neurosurgical case is not available to serve as matches in subsequent trials. This process was repeated until a match was found for every orthopaedic case. All descriptive and comparative statistics in this study were performed using SAS software, version 9.2 (SAS Institute, Cary, NC).

3. Results

The 2011-2014 NSQIP database yielded 17,967 ACDF procedures, of which 4,611 (25.6%) were orthopaedic and 13,356 (74.3%) were neurosurgical. More patients within the neurosurgical cohort were ASA ≥3 (40.8% versus 35.9%, p < 0.00001) with a greater burden of diabetes (15.2% versus 13.6%, p < 0.0074). All other demographic variables and comorbidities were comparable (Table 1).

Table 1.

Patient demographics by surgeon specialty.

Characteristic	Orthopaedic Surgery Value N (%)	Patients for whom characteristic was determined, (N)	Neurosurgery Value N (%)	Patients for whom characteristic was determined, (N)	Chi Square or T-test p-Value
Age, mean (SD)	53.3(11.3)	4611	53.8(11.6)	13356	0.0143*
Sex (% male)	2287(49.6)	4611	6494(48.6)	13352	0.2599
BMI ≥ 30	2075(45.2)	4588	5876(44.2)	13290	0.2339
ASA ≥ 3	1654(35.9)	4609	5443(40.8)	13333	<0.00001*
Diabetes Mellitus	628(13.6)	4611	2036(15.2)	13356	0.0074*
History of Cardiac Disease	41(0.9)	4611	110(0.8)	13356	0.6741

*Significant value.

† SD – standard deviation, BMI – body mass index; ASA – American Society of Anesthesiologists.

There was a trend towards longer operative times in patients undergoing ACDF with neurosurgery (p = 0.0609). The rates of overall and major systemic complications within the thirty-day postoperative period were also comparable (Table 2). There was no significant difference in length of hospital stay, rate of ventilator dependency >48 h, and hospitalization past 30 days. However, significantly more patients were discharged to extended care facilities following neurosurgical ACDF (5.3% versus 4.2%, p = 0.0047). Finally, rates of 30-day readmission and unplanned reoperation were similar in both groups.

Table 2.

Complications and outcomes by surgical specialty.

Characteristic	Orthopaedic Surgery Value N (%)	Patients for whom characteristic was determined, (N)	Neurosurgery Value N (%)	Patients for whom characteristic was determined, (N)	Chi Square or T-test p-Value
Operative Time, min, mean ± SD	124.0 ± 72.8	4611	126.3 ± 69.4	13355	0.0609
Overall Complications	113(2.5)	4611	372(2.8)	13356	0.2268
Major Systemic Complications	61(1.3)	-	178(1.3)	-	0.9600
Pulmonary Embolism	8(0.2)	-	21(0.2)	-	0.8125
Unplanned Intubation	28(0.6)	-	75(0.6)	-	0.7231
Ventilator > 48 h	20(0.4)	-	64(0.5)	-	0.6965
Postoperative Sepsis/Septic Shock	15(0.3)	-	59(0.4)	-	0.2871
Stroke/Cerebrovascular Accident	5(0.1)	-	11(0.1)	-	0.6088
Acute Renal Failure	0(0)	-	8(0.1)	-	0.0965
Cardiac Arrest Requiring CPR	8(0.2)	-	21(0.2)	-	0.8125
Myocardial Infarction	7(0.2)	-	16(0.1)	-	0.6001
Coma	0(0)	-	2(0.01)	-	0.406
Minor Systemic Complications	58(1.3)	-	197(1.5)	-	0.2825
Urinary Tract Infection	16(0.4)	-	83(0.6)	-	0.03
Deep Venous Thrombosis	17(0.4)	-	35(0.3)	-	0.2452
Pneumonia	28(0.6)	-	96(0.7)	-	0.4303
Progressive Renal Insufficiency	2(0.04)	-	2(0.01)	-	0.2651
Major Local Complications	11(0.2)	-	35(0.3)	-	0.7855
Deep Wound Infection/Organ or Space SSI	11(0.2)	-	33(0.3)	-	0.9196
Peripheral Nerve Injury	0(0)	-	1(0.01)	-	0.5568
Graft/Prosthesis Failure	0(0)	-	1(0.01)	-	0.5568
Minor Local Complications	12(0.3)	-	45(0.3)	-	0.4247
Superficial Wound Infection	12(0.3)	-	39(0.3)	-	0.7267
Wound Disruption	0(0)	-	9(0.1)	-	0.0779
Mortality or Major Complication	70(1.5)	-	208(1.6)	-	0.8523
Hospital Length of Stay, mean (SD)	1.9(6.2)	4609	2.0(5.8)	13351	0.5857
Readmission	128(2.8)	4547	418(3.2)	13180	0.2303
On Ventilator >48hrs	20(0.4)	4611	64(0.5)	13356	0.6965
Hospital Discharge Destination
Home	4411(95.8)	4607	12621(94.7)	13329	0.0047*
Extended Care Facility	196(4.2)	-	708(5.3)	-	-
Still in Hospital >30 days	3(0.1)	4106	23(0.2)	11757	0.0946
Post-op Death >30 days if in acute care	0(0)	3	0(0)	23	-
Unplanned re-operation	59(1.3)	4586	194(1.5)	13305	0.3961

*Significant value.

SD – standard deviation.

A 1:1 propensity match was subsequently performed between surgical specialties that yielded 4,582 well-matched pairs (Table 3). After matching, patient demographics of the adjusted analysis were more comparable. There was no significant change in mean operative time, overall and major systemic complication rates, and hospital length of stay for patients undergoing ACDF procedures.

Table 3.

Matched pairs based on propensity score.

Characteristic	Orthopaedic Surgery Value N (%)	Patients for whom characteristic was determined, (N)	Neurosurgery Value N (%)	Patients for whom characteristic was determined, (N)	McNemar Test or Paired T-test p-Value
Age, mean (SD)	53.3(11.3)	4581	53.3(11.2)	4581	0.6769
Operative Time, min, mean ± SD	124.1 ± 72.8	-	126.8 ± 71.5	-	0.0784
Overall Complications	112(2.4)	4582	118(2.6)	4582	0.7373
Major Systemic Complications	60(1.3)	-	55(1.2)	-	0.7018
Mortality	11(0.24)	-	11(0.24)	-	1.0000
Hospital Length of Stay, mean (SD)	1.9(6.2)	4578	2.0(6.4)	4578	0.8567

*Significant value.

SD – standard deviation.

4. Discussion

Since the 1950s, cervical fusion procedures have become increasingly accepted for addressing degenerative pathology.^1,2,22 ACDF is amongst the most frequently performed elective cervical spine surgeries, and therefore serves as an excellent basis by which to investigate differences in outcomes between orthopaedists and neurosurgeons. Spine surgery in and of itself poses significant stress to the patient, and it is imperative that this is not propagated by concerns over which specialty-trained surgeon they should select. In the present study, we found that although neurosurgeons performed nearly three-times the number of ACDF procedures, there were no significant differences in multiple outcome metrics.

Previous literature evaluating differences between orthopaedic and neurosurgical specialties are limited and studies have yielded conflicting results.^14,15 Seicean et al. utilized an earlier iteration of the NSQIP (2006-2012) to analyze differences in outcomes following elective spinal surgeries performed by either orthopaedic or neurosurgeons.¹⁴ The authors reported that patients who underwent orthopaedic procedures were significantly more likely to have a prolonged length of stay, higher rate of postoperative complications, and necessitate discharge to extended care facilities. However, the differences were marginal and the cohort in question was heterogeneous, precluding evaluation of specific procedures such as ACDF. Using the same NSQIP dataset (2006-2012), Minhas et al. investigated specialty training-based differences in 30-day outcomes following isolated, single-level ACDF procedures.¹⁵ Of the 1,944 patients included in the analysis, the authors reported a significantly higher proportion of neurosurgical procedures (80.1% versus 19.9%). While patients treated by neurosurgeons had a higher comorbidity burden, there was ultimately no difference in complications between the two cohorts. Similar to this study, we evaluated a more homogenous population of isolated ACDF procedures with a later iteration of the NSQIP dataset (2011-2014). This allowed for inclusion of nearly 10-times the number of patients and therefore a more appropriately powered analysis of differences in outcomes between surgical specialties. In contrast to the findings of Seicean et al. in our analysis of over 17,000 ACDF procedures, we found that the neurosurgical cohort experienced a trend towards longer operative times with significantly more patients being discharged to extended care facilities. One possible explanation for these findings is that neurosurgeons typically take more trauma spine call and thus may operate on sicker, more complex patients. Our analysis also includes a more recent and narrower dataset (2011-2014), which is important considering the updated addition of NSQIP metrics such as 30-day readmission and ongoing improvements in quality control. Furthermore, continued advancements in imaging technology, surgical techniques, and patient expectations contribute to the constantly evolving training environments of each specialty. Our results support the finding that for primary ACDF, surgeon specialty training does not significantly affect complications.

The significance of specialty training of spinal surgeons remains a point of contention. Currently, in the United States, spine surgeon training consists of either a neurosurgical residency or orthopaedic residency followed by an additional year of subspecialty fellowship training.^23,24 The Accreditation Council for Graduate Medical Education (ACGME) has mandated specific and largely dissimilar requirements for orthopaedic and neurosurgical residencies. A neurosurgical residency generally comprises 84 months, during which residents have greater minimum requirements for spine surgery exposure and thereby dedicate more time on average to spine-specific training than orthopaedic residents.²⁵ Most neurosurgeons enter independent practice thereafter without receiving additional fellowship training. Orthopaedic surgical residency, in contrast, most often consists of 60 months of training within a multitude of subspecialties.²⁴ Orthopaedic residency programs introduce the fundamentals of spine surgery to serve as a foundation for additional subspecialty training, while neurosurgical residency programs attempt to provide the skills to practice spinal surgery independently.²⁶ It is important to note that orthopaedic trainees are exposed to more instrumentation and osseous fixation techniques throughout their residency training than neurological trainees.²⁷ Presently, the ACGME requires a minimum of only 15 decompression and/or posterior fusion cases to be completed during orthopaedic residency, and as a result essentially all orthopaedic surgeons who plan to practice spine surgery go on to pursue a spine fellowship prior to entering independent practice.²⁸ Highlighting this difference in training, a recent survey found that 87.4% of orthopaedic residents and 41.5% of neurosurgical residents planned on pursuing additional training prior to entering independent practice.¹³ These differences in surgical case volume, exposure, and overall length of training contribute to varying degrees of technical skill and decision-making and may ultimately impact outcomes.29, 30, 31 Nonetheless, our results suggest that differences in subspecialty training may not significantly affect outcome metrics of interest for more common procedures such as ACDF.

The findings of this study should be considered in light of several methodological limitations. The retrospective nature of this study portends inherent bias. In addition, the NSQIP does not contain information regarding potential variables such as hospital volume, geographic location, quality of life measures, or skill level of individual surgeons. Our statistical models are limited to the variables recorded in the dataset; therefore the potential for residual confounding cannot be completely eliminated. Rates of readmission by operative indication and surgical technique for ACDF cases could not be stratified because they were collectively classified under a common CPT code during the study period. Protocol limitations furthermore prevent detection of the possible involvement of multiple surgical specialties in a given case. Importantly, the thirty-day duration of follow-up inhibits detection of subsequent complications or reoperations that may occur outside this parameter. Despite these limitations, the NSQIP represents a previously validated, large, multi-institutional repository of prospectively collected outcomes data that serves to aid in surgical quality improvement. Multiple supervisory quality-control measures further ensure integrity and consistency of the database by limiting reporting errors.

5. Conclusion

Differences in outcome metrics following ACDF when performed primarily by either orthopaedics or neurosurgery is important in the decision making process for the spine patient. The results from the present analysis of 17,967 cases suggest that ACDF is successfully performed by either specialty without concern that substantial outcome differences exist when selecting one surgeon over the other.

Declaration of competing interest

The authors have no conflict of interest. We do not have any financial or personal relationships to disclose.