Key points

Introduction

Cost and value (defined as the quality or outcomes of care compared with cost) are increasingly important components of health care. Despite a wealth of CEAs in many areas of medicine, there has been little research addressing the cost of neurosurgical procedures until recently. This is particularly problematic because this specialty represents one of the most expensive areas in medicine. According to the Centers for Disease Control and Prevention, there were approximately 1.2 million neurosurgical procedures performed in the United States in 2010. The cost of lumbar laminectomies alone exceeded $2 billion, and spinal fusions cost $12.8 billion nationwide in 2011.

This article first discusses the general principles of CEAs, then reviews the cost-related research that has been done to date in the neurosurgical subspecialties, primarily spine and also trauma, functional, vascular, pediatric, and tumor neurosurgery. Finally, the need for standardization of cost and cost-effectiveness metrics within neurosurgery is highlighted and an easy-to-use set of metrics to guide future research in neurosurgical cost-effectiveness is defined.

Introduction

Cost and value (defined as the quality or outcomes of care compared with cost) are increasingly important components of health care. Despite a wealth of CEAs in many areas of medicine, there has been little research addressing the cost of neurosurgical procedures until recently. This is particularly problematic because this specialty represents one of the most expensive areas in medicine. According to the Centers for Disease Control and Prevention, there were approximately 1.2 million neurosurgical procedures performed in the United States in 2010. The cost of lumbar laminectomies alone exceeded $2 billion, and spinal fusions cost $12.8 billion nationwide in 2011.

This article first discusses the general principles of CEAs, then reviews the cost-related research that has been done to date in the neurosurgical subspecialties, primarily spine and also trauma, functional, vascular, pediatric, and tumor neurosurgery. Finally, the need for standardization of cost and cost-effectiveness metrics within neurosurgery is highlighted and an easy-to-use set of metrics to guide future research in neurosurgical cost-effectiveness is defined.

Principles of cost-effectiveness analyses

A CEA is a type of economic analysis that compares the costs and health outcomes of 2 or more courses of action. CEAs are often expressed in terms of a ratio of cost per health gain. The most commonly used health outcomes measure in the United States and Europe is QALYs. A QALY reflects both the quantity and quality of the years gained by a medical intervention, and is equal to time (years) × quality (ie, utility). Health utility is on a scale from 0 to 1, with 0 indicating death and 1 representing perfect health. Direct methods to estimate health state utility include time tradeoff, standard gamble, and visual analog scale. Indirect methods include the Health Utility Index, EuroQoL–5 Dimension (EQ-5D), and Short Form–6 Dimension (SF-6D). A single year spent in perfect health yields 1 QALY, and effective medical interventions increase QALYs. To compare 2 interventions (eg, treatments A and B), an ICER, which equals (cost of B – cost of A)/(QALYS with B – QALYs with A), is calculated. The use of ICERs enables the cost of achieving a certain benefit to be compared with similar ratios calculated for other health interventions, providing a broader context in which to make judgments about the value for money of a particular health intervention. In the United States, ICERs less than $150,000 are typically considered cost effective, because this represents 2 times the gross domestic product per capita.

A cost-utility analysis is a specific type of CEA that uses health utilities expressed as QALYs (described previously). CEAs can also include other health outcomes, such as cost per death averted or added year of life. A cost-benefit analysis, distinct from a CEA, assigns a monetary value to health outcomes, usually based on a population’s “willingness to pay” for those outcomes. Thus, it calculates the net monetary cost or savings of an intervention. It is used less frequently than CEAs in medicine.

A rigorous CEA must specify its cost methods. Costs include both direct costs (eg, resources consumed by the surgical procedure, such as surgical implants and hospital stay, and the costs of future medical care) and time costs (eg, due to loss of productivity from the morbidity of a surgical procedure). In the literature, hospital-allowed charges (ie, what the hospital is paid by the insurance company) are often used as a proxy for direct cost. Importantly, crude (billed) hospital charges can bear little resemblance to economic cost ; and use of hospital charges as a proxy for cost may lead researchers to draw unwarranted conclusions. The best measure of cost is actual resource utilization, which can be difficult to calculate but is available via some hospital cost-accounting systems. Many articles in the literature are forced to use insurance payments, specifically the Centers for Medicare and Medicaid Services reimbursement values for specific diagnosis-related group and current procedural terminology codes, as estimates for cost.

Several additional aspects of CEA methods should also be reported in each study. These include the analytical time period and perspective (eg, that of society or health care payers), the discount rate, the type of sensitivity or uncertainty analysis performed, and the selected cost-effectiveness threshold (if used). All these criteria are reported in the Cost-Effectiveness Analysis Registry, which is a comprehensive database of 4007 cost-utility analyses that have been assessed by reviewers with training in cost-effectiveness and decision analysis. The CEA model structure and input values must be transparent and thoroughly documented and justified, typically with some detail in online supplemental documents.

Cost-effectiveness analyses in neurosurgery

A comprehensive PubMed search for “cost-effectiveness” and “neurosurgery” had 691 hits, although only a small subset of these results were true CEAs. A more refined search (“cost-effectiveness” [ti] “cost utility” [ti] neurosurgery) helped narrow the list to 140 articles. A search of the Cost-Effectiveness Analysis Registry (search terms, “neurosurgery” and “spine”) revealed fewer than 50 verified cost-utility analyses in the field of neurosurgery up to early 2013, a majority of which are in the subspecialty of spine. Admittedly, there has been an increased interest in this area recently, with a significant rise in the number of neurosurgery cost-effectiveness studies published over the past 2 years. A majority of purported cost-effectiveness neurosurgery studies, however, do not adhere to the CEA methodology described previously. Many of these are actually cost comparison (ie, descriptive comparisons of cost differences) rather than cost-effectiveness studies. They also have several limitations, including inconsistent use of cost methods (direct vs indirect costs, charges vs payments), variable outcome measures, and potentially noncomparable data sets (ranging from large national databases, such as the Nationwide Inpatient Sample database, to small, single-institution series).

Spine: the leader in neurosurgery cost-effectiveness

Driven largely by the high costs of their procedures and insurance companies’ demands for justification of their interventions, spine surgeons were among the first neurosurgeons to enter the cost-effectiveness field. One of the earliest studies, published in 2008, showed the cost-effectiveness of lumbar laminectomy, compared with nonoperative treatment, for lumbar disc herniation at 2 years (ICER $69,403). Using the same Spine Patient Outcomes Research Trial data, this research group also found that lumbar laminectomy was a cost-effective treatment option compared with nonoperative treatment for spinal stenosis with and without degenerative spondylolisthesis at 2 years (ICER <$150,000). These findings were supported by a study in 2010 showing that lumbar laminectomy was more effective than nonsurgical care or X-STOP Interspinous Process Decompression System for treatment of symptomatic lumbar spinal stenosis (ICER = $28,256 for laminectomy compared with conservative management at 4-year follow-up). More recently, several studies have looked at the cost-effectiveness of decompression alone versus instrumented fusion for grade I L4/5 spondylolisthesis ($56,610/QALY for decompression alone vs >$100,000/QALY gained for decompression with various types of fusion), the cost/QALY gained for transforaminal lumbar interbody fusion (TLIF) for grade I degenerative spondylolisthesis (2-year cost of $42,854 per QALY gained), and the cost-effectiveness of revision surgery for same-level recurrent lumbar stenosis and adjacent-segment disease (2-year cost of $80,594 per QALY gained). Much of this work comes from the same research group (led by Matthew McGirt, MD, formerly of Vanderbilt University) performing analysis on a small set of patients at a single institution.

Similar to this body of work on degenerative lumbar disease, several groups have investigated the cost-effectiveness of different surgical treatment options for cervical spine disease, metastatic spine tumors, and adult spinal deformity. Researchers have also examined the cost-effectiveness of specific spinal implants, such as bone morphogenic protein ($136,207/QALY gained), femoral head allografts, and polyetheretheketone anterior cervical cages (>$100,000/QALY gained), as well as the cost-effectiveness of certain intraoperative techniques like neurophysiological monitoring and O-arm confirmation of lumbar pedicle screw placement. With the increasing popularity of minimally invasive spine (MIS) surgery, many recent spine CEAs have focused on determining the cost-effectiveness of newer minimally invasive techniques, such as the minimally invasive TLIF or tubular discectomy, compared with the traditional open approaches. Results for MIS surgery have been mixed, with some studies suggesting that minimally invasive approaches are cost effective but others showing equivalent cost-effectiveness for minimally invasive and open approaches.

Despite the growing number of studies on spinal surgery cost-effectiveness, there are many limitations of this work. First, many of these studies are simply cost descriptions rather than true CEAs, as discussed previously. These studies use widely different measures of cost, from hospital-based charges to Medicare reimbursement rates, which can substantially influence the results of a cost-effectiveness study, and are discussed in further detail later. Finally, most of these are retrospective single-institution studies. To make systems-level treatment recommendations, more robust CEAs from prospective multicenter studies are needed, such as the currently underway Verbiest trial in the Netherlands comparing operative with nonoperative treatment of neurogenic claudication due to lumbar stenosis and the Netherlands Cervical Kinematics double-blind randomized multicenter study comparing the cost-effectiveness of anterior cervical discectomy with and without interbody fusion and arthroplasty.

Cost-effectiveness analyses in other areas of neurosurgery: moving beyond spine

Despite the growing literature focused on cost-effectiveness in spinal surgery, there are only a handful of articles addressing cost issues in other neurosurgical subspecialties. In the authors’ comprehensive literature review, 7 articles were identified addressing cost-effectiveness in neurosurgical trauma. Three studies addressed the cost-effectiveness of decompressive hemicraniectomy for severe traumatic brain injury (TBI): 1 from the United States and another from Europe reported that decompressive hemicraniectomy was a cost-effective treatment option for patients with severe TBI (even if they were >80 years old [eg, €17,900/QALY gained]), but a study from western Australia concluded that surgery was not a cost-effective option for patients with severe TBI when the predicted risk of an unfavorable outcome was greater than 80% ($682,000/QALY). Another study found that surgical decompressive hemicraniectomy was more cost-effective than a barbiturate coma for refractory intracranial hypertension (ICER = $9,565). Finally, 2 studies examined the cost-effectiveness of radiographic imaging for traumatic minor head injury in adults and in children.

In the realm of functional neurosurgery, 3 recent studies examined the cost-effectiveness of microvascular decompression for trigeminal neuralgia compared with radiosurgery and percutaneous rhizotomy ($4,931/QALY for surgery vs $7,768/QALY for radiosurgery and $602/QALY for percutaneous rhizotomy). None of these studies compared microvascular decompression with medical therapy, which is an important target for future work. A recent study from Hong Kong looked at the ICER of deep brain stimulation versus medical therapy for Parkinson disease (ICER = $123,110 at 1 year and $62,846 at 2 years). This analysis did not calculate cost-effectiveness over patient expected lifetime, however. Two international studies also showed cost-effectiveness of surgical treatment of epilepsy compared with continued medical therapy (ICER = $25,020 to $69,451 Canadian dollars), which promises to be an exciting area of cost-effectiveness research for the functional neurosurgeon.

Only 3 published works have applied cost-effectiveness techniques to pediatric neurosurgery, with 1 group performing a cost comparison analysis for endoscopic-assisted craniectomy versus open cranial vault remodeling for sagittal synostosis, another looking at the cost-effectiveness of endoscopic third ventriculostomy versus shunt for hydrocephalus, and another looking at the costs and benefits of neurosurgical intervention for infants with hydrocephalus in sub-Saharan Africa ($59 to $126/disability-adjusted life year averted). The cost-effectiveness literature in vascular neurosurgery is similarly sparse ; and studies comparing the cost of surgical clipping versus endovascular treatment of ruptured aneurysms provide contradictory results.

Neurosurgical oncology cost-effectiveness analyses

As in trauma, functional, pediatric, and vascular neurosurgery, there is a paucity of literature addressing the cost of neurosurgical oncology (ie, brain tumor treatment), and a majority of these articles are from outside the United States. The authors’ research group, therefore, has focused efforts on performing the first rigorous CEAs with decision-tree analyses for the management of benign brain tumors, including vestibular schwannomas, prolactinomas (prolactin-secreting pituitary tumors), and meningiomas. Their vestibular schwannoma CEA using direct hospital cost data found that surgery is a cost-effective alternative to radiation when patients are diagnosed with a vestibular schwannoma at less than 45 years old (ICER <$150,000). For vestibular schwannoma patients greater than or equal to 45 years old, radiation is the most cost-effective treatment option. In a prolactinoma CEA analysis, despite higher up-front surgical costs, surgery was the less expensive treatment option over a patient’s expected lifespan ($428 per 1% reduction in serum prolactin level for surgery vs $921 for bromocriptine and $1621 for cabergoline) because medically treated patients must often remain on either bromocriptine or cabergoline indefinitely. In the authors’ CEA model, surgery is a cost-effective alternative to cabergoline and bromocriptine at all ages of diagnosis less than 80 years old.

Discussion

This article provides a brief summary of the basic principles of CEAs as well as a review of the cost-related research in all subspecialties of neurosurgery, including spine, trauma, functional, pediatrics, vascular, and tumor neurosurgery. The limitations of much of the neurosurgery cost-effectiveness work that has been performed to date are emphasized. Many of the studies discussed in this article are not true CEAs but rather cost studies. Even the studies that are CEAs often use different cost metrics, variable outcome measures, and unreliable data sets, making it difficult to draw definitive conclusions from them. A particularly nice study showed how the choice of cost method (hospital-based cost analysis using charges multiplied by cost-to-charge ratios vs Medicare reimbursements) substantially influenced the final results of a cervical spine surgery CEA. Medicare reimbursements may underestimate real cost whereas hospital charges can grossly overestimate true costs.

It is, therefore, of utmost important for this field to establish guidelines for cost and CEA methods and reporting so that results between studies can be appropriately compared. The authors propose that every neurosurgery cost-effectiveness article be held to the basic standards of CEA analysis used by the Cost-Effectiveness Analysis Registry. More specifically, every neurosurgery cost-utility analysis should have correctly calculated QALYs and ICERs and should explicitly report the analytical time period and analytical perspective (eg, from the societal or health care payer perspective), the currency used, the discount rate, the type of sensitivity or uncertainty analysis performed, and the selected cost-effectiveness threshold (if used). In addition, model design and input values should be sufficiently described to permit replication and comparison with other analyses that might use a different structure or input values.

Finally, to make systems-level treatment recommendations, greater generalizability is needed, building the CEAs on prospective multicenter studies, a few examples of which are already under way in Europe. In the United States, the authors hope that such studies will become possible with further developments of the new national clinical databases, such as the National Neurosurgery Quality and Outcomes Database.