Vascular imaging of the spine in the US Medicare population: Catheter and MR angiography volumes from 2004 to 2016

Introduction

The vascular anatomy of the spine is highly intricate and complex.^1,2 Spinal catheter angiography (CA) is considered the gold standard for the diagnosis and classification of spinal vascular lesions, and is often a prelude to intervention. ³ However, advanced computed tomography angiography (CTA) and magnetic resonance angiography (MRA) techniques now provide improved non-invasive evaluation of spinal vascular anatomy and pathology.^4–6 A robust understanding of spinal vasculature, its variant anatomy, vascular pathophysiology, and pathologic mimics is necessary in the planning and treatment of spinal disease by endovascular techniques. As such, it is important for the endovascular spinal neurointerventionalist to have a robust background training in the neurosciences. Disciplines of the neurovascular triad (namely neuroradiology, neurosurgery, and neurology) by virtue of their training best fit the profile of an endovascular spinal neurointerventionalist. Prior studies have shown that radiologists, neurosurgeons, neurologists, cardiologists, and vascular surgeons are amongst the specialties that are heavily involved in endovascular procedures of the head and neck.^7,8 However, not much is known about the major specialties involved in endovascular neurointerventions of the spine. The purpose of our study is two-fold: firstly, to investigate the physician specialties that are involved in endovascular spinal neurointerventions. As quality and patient safety concerns are continuously scrutinized, this information will be of interest to various stakeholders, including current and aspiring neurointerventionalists, government agencies, policy makers, payers, and third-party contractors. Secondly, we sought to understand the changing patterns in the utilization of spinal CA especially with the advent of more modern non-invasive spinal angiographic techniques, by comparing volumes of spinal CA with MRA.

Materials and methods

Results

The volume of spinal CA procedures performed in the United States (US) Medicare fee-for-service population was 4758 in 2004, peaked at 6869 in 2012, and decreased slightly to 6656 in 2016. Overall, the volume of spinal CA increased by 40% from 2004 to 2016 (Figure 1). Specialty-specific analysis of the volumes demonstrated that radiologists performed most of these procedures, conducting 3736 procedures in 2016 (56.1%). Neurosurgeons were the second-largest specialty involved in spinal CA, performing 2456 procedures in 2016 (36.9%), followed by neurologists, who performed 346 procedures in 2016 (5.2%) (Figure 2). Collectively, the remaining specialties including vascular surgeons and cardiologist performed 118 procedures in 2016 (1.8%).

OPEN IN VIEWER

Figure 1.

Overall utilization of spinal catheter angiography in the Medicare fee-for-service population from 2004 to 2016.

OPEN IN VIEWER

Figure 2.

Spinal catheter angiography volumes analyzed according to physician specialty in the Medicare fee-for -service population from 2004 to 2016. RADS: Radiology; CARDS: Cardiology; NSURG: Neurosurgery; NEU: Neurology; VASC SURG: Vascular surgery; OTHER: Other physicians.

Over the 12-year period, the proportion of procedures performed by radiologists steadily decreased from 88.0% in 2004 to 56.1% in 2016 (Figure 2). Conversely, the proportion of procedures carried out by neurosurgeons and neurologists increased, from 9.9% to 36.9% and 0.1% to 5.2% respectively from 2004 to 2016. The proportion of procedures by the remaining specialties collectively remained largely unchanged except for a few fluctuations: 1.9% in 2004, with a peak in 2008 of 7.4%, then a subsequent decrease to 1.8% in 2016 (Figure 2).

Further analysis of the number of procedures per physician over a 9-year period revealed a decrease for radiologists from 163 procedures per 1000 physicians in 2007 to 135 procedures per 1000 physicians in 2015. A 4.5-fold decrease was seen for vascular surgeons from 18 procedures per 1000 physicians to four procedures per 1000 physicians over this period. Conversely, there was about three-fold increase for neurosurgeons (109 procedures per 1000 physicians to 347 procedures per 1000 physicians) and four-fold increase for neurologists (four procedures per 1000 physicians to 40 procedures per 1000 physicians) respectively from 2007 to 2015 (Figure 3). Trends for cardiologists remained steady, moving from four procedures per 1000 physicians to three procedures per 1000 physicians. Trend lines fitted to specialties with the highest procedure-physician ratios revealed a linear curve for radiologists and neurologists, but a logarithmic curve for neurosurgeons.

OPEN IN VIEWER

Figure 3.

Number of spinal catheter angiography procedures per 1000 physician (procedure/physician ratio) by specialty group from 2007 to 2015. Trends lines are best fitted to top three specialties, namely neurosurgery, radiology and neurology.

The volume of spinal MRA procedures performed in the US Medicare fee-for-service population fluctuated between 0 and 1 from 2004 to 2009, then precipitously increased to 40 in 2010, peaked at 133 in 2011, and slowly declined to 81 in 2016. The volume of spinal MRA procedures increased by 8000% from 2004 to 2016 (Figure 4). Throughout the analyzed period, radiologists were involved in the majority of spinal MRA procedures, interpreting no less than 93% of studies during any given year (Figure 4).

OPEN IN VIEWER

Figure 4.

Spinal magnetic resonance angiography volumes analyzed according to physician specialty in the Medicare fee-for-service population from 2004 to 2016.

Discussion

Our study demonstrated an overall slow progressive increase in the number of spinal CA procedures performed since 2004 with minor fluctuations and an increase in spinal MRA procedures, particularly after 2009. Despite increases in the utilization of both procedures, the vast majority of spinal vascular pathology is still evaluated by CA (6656 CA versus 81 MRA procedures in 2016). The reason for the low utilization rate of spinal MRA is unclear. Advances in MRA techniques for spinal vasculature now provide reasonable screening evaluation of spinal vascular lesions, particularly for arteriovenous fistulas.^5,6 MRA imaging techniques are rarely performed in isolation, but as an adjunct to conventional MR imaging.^5,10 Before performing digital subtraction angiography (DSA), MRA imaging can be used to predict the location of spinal fistulas, which would reduce the amount of time and radiation used during the DSA procedure. However, it is likely that in practice, the performance of spinal MRA does not preclude diagnostic spinal CA, which commonly precedes the embolization. It is also possible that many clinicians recognize spinal CA to be the gold standard and are more familiar with it. Thus, spinal MRA techniques—despite their advances—are most commonly familiar to radiologists and have not yet been assimilated into many clinical imaging algorithms for patients with spinal vascular pathology. This is consistent with our sub-analysis of the volumes of spinal MRA by physician specialty, where radiologists were involved in > 93% of all spinal MRA studies in the Medicare fee-for-service population over the studied period (Figure 4).

From our analysis, we found that the specialties that comprise the neurovascular triad are the largest performing specialties for spinal CA in the Medicare fee-for-service population, with radiologists providing the majority of procedures compared with neurosurgeons and neurologists. However, the proportion of procedures by radiologists decreased from 88.0% in 2004 to 56.1% in 2016. In contrast, the proportion of procedures by neurosurgeons and neurologists increased from 9.9% to 36.9%, and from 0.1% to 5.2% respectively from 2004 to 2016 (Figure 2), despite the gradual steady increase in total volume of spinal CA procedures by 1.4-fold over the studied period (Figure 1). Along the same lines, from 2007 to 2015, the procedure-physician ratio trends indicated a gradual decrease for radiologist, and a gradual increase for neurologists, but a substantial increase for neurosurgeons (Figure 3). These trends are not unexpected, and closely match the physician specialty distribution in the performance of endovascular procedures of the head and neck. In a study by Cox et al., despite a six-fold increase in neurointerventional procedures of the head and neck, there was a decrease in the proportion of procedures performed by radiologists from 80% in 2000 to 30% in 2013. Conversely, there was an increase in the proportion of procedures performed by neurosurgeons (10% to 19%), and neurologists (0.04% to 6.3%) from 2000 to 2013. ⁸ The reasons for this gradual shift in specialty distribution are in part due to the development of the endovascular neurosurgical and interventional neurology subspecialties within the neurosurgery and neurology fields respectively, which have served as alternative channels to performing these procedures. Of note, there was a sharp increase in the volume of spinal CA procedures performed by neurosurgeons between 2014 (22.8%) and 2016 (36.9%). This increase was virtually matched by a decrease in volume of spinal CA procedures performed by radiologists (66.2% in 2014 to 56.1% in 2016). Although there remain concerns about turf wars within the neurovascular triad, the positive outcomes of several trials featuring the benefit of endovascular stroke interventions^11–17 and subsequently FDA approval of multiple mechanical thrombectomy devices,^18,19 in combination with the rise in demand for acute ischemic stroke interventions, ⁷ has resulted in a shift in the prevailing ethos to one of collaboration to meet the demand. ²⁰ This is not inconsistent with our findings, which suggest a gradual flattening of the procedure-physician curve for neurosurgeons after 2010 (Figure 3). Consequently, we believe the perceived turf wars within the neurovascular triad are becoming less of a concern, and the main focus now is on providing standards for training and credentialing to ensure the quality of graduating neurointerventionalists, and reduce the number of inadequately trained, non-neurovascular specialists entering clinical practice who seek to perform these procedures.^21–23

Unlike endovascular procedures of the head and neck, where vascular surgeons and cardiologists perform a significant volume, particularly with extracranial procedures,^7,8 these specialties play a negligible role in spinal CA. One possible reason for this is that spinal vascular anatomy, relative to cervicocerebral anatomy, is more intricate and complex with several anatomic variants, ¹ and is unfamiliar to interventionalists without a robust background in neuroscience. There is also the potential for significant morbidity and even mortality if complications were to emerge, with potentially significant medicolegal consequences particularly if the performing physician is found to be operating outside the traditional boundaries of their specialty. This observed partiality of spinal CA procedures to the neurovascular triad may be in the best interests of patients who need vascular interventions of the spine as their background training in the neurosciences makes them best qualified to manage and treat vascular pathology of the spine.

There are some limitations to this study. First, our data were obtained from the Medicare database, which although covering approximately one-sixth of the US population, does not include patients under 65 years of age. Despite this, the main objective of the study was to observe trends and physician specialty involvement over the studied duration. We speculate that these trends are unlikely to significantly vary, even if data from patients under 65 years of age were included in our analysis. Secondly, because there are no distinct CPT codes for spinal tumor and vascular embolizations, spinal CA was used as a surrogate measure for the volume of spinal procedures. However, because spinal CA commonly precedes the embolization procedure itself, omission of the embolization procedure is unlikely to significantly alter the utilization trends or physician specialty distribution. Similarly, because the specific number of providers within each specialty performing spinal CA procedures could not be obtained from the PSPSMF database, we utilized total volumes of practicing physicians within the selected specialties as a surrogate measure. Accordingly, no assertions can be made about the absolute procedure-physician ratios, which carry far less significance compared with the trends that these values portray. Third, we have postulated that the neurovascular triad (neurosurgery, neurology, and neuroradiology) are the specialties that are best suited for endovascular spinal neurointerventions by virtue of their background training. Although we believe this to be a reasonable postulation, we recognize that the presence of morbidity and mortality data would strengthen this position.

Conclusion

Our results show that spinal CA volumes continue to rise in the Medicare fee-for-service population, and most of the procedures are being performed by the neurovascular triad, which we believe to be in the best interest of patients with pathologies requiring endovascular spinal neurointerventions. Although spinal MRA volumes have started to increase, they comprise only a small fraction of studies performed for vascular evaluation of the spine, possibly owing to lack of familiarity with the procedure among non-radiologist physicians.