Economic and Outcomes Analysis of Recalcitrant Cervical Radiculopathy

Is Nonsurgical Management or Surgery More Cost-Effective?

Jeffrey A. Rihn, MD; Suneel Bhat, MD; Jonathan Grauer, MD; James Harrop, MD; Zoher Ghogawala, MD; Alexander R. Vaccaro, MD, PhD; Alan S. Hilibrand, MD


J Am Acad Orthop Surg. 2019;27(14):533-540. 

In This Article


Model Design

The presented model consisted of two primary treatment arms. Patients with cervical radiculopathy failing 6 weeks of nonsurgical management could either undergo invasive nonsurgical management with cervical epidural injections and physical therapy and potentially surgery for failed treatment or proceed directly to one- or two-level ACDF.

In the model, patients initially managed with surgery either had successful resolution of their symptoms or went on to develop symptomatic pseudarthrosis, infection, hematoma requiring revision surgery, nerve root injury/persistent radiculopathy, or spinal cord injury. The patients who had revision surgery after undergoing a revision ACDF showed successful resolution of symptoms or chronic dysfunction. Patients with successful surgery had an annual risk of adjacent segment disease (ASD) each year they survived, and if they developed persistent symptoms, underwent revision ACDF with the same associated postoperative risks and complications. The complications of nerve root injury, failure of improvement of symptoms, or hematoma did not receive further treatment and remained in their respective states until death.

Patients initially managed by epidural injections underwent a series of three injections over a 6-week period along with a concurrent course of physical therapy; they either had successful improvement of symptoms, conversion to ACDF because of the failure of treatment with the associated complications, and outcomes as described in the surgical group or were removed from the model because of injection-related complications of nerve root or spinal cord injury. Those who had successful improvement of their symptoms had an annual risk of same segment or ASD each time they survived into the next year. If symptoms developed, the patients underwent the same treatment approach as primary disease, and subsequently after accruing associated costs and outcomes, they remained in that state until death. The complications of nerve root injury, failure of improvement of symptoms after conversion to ACDF, or hematoma did not receive further treatment and remained in their respective states until death.

Patients were removed from the model at age-specific mortality rates or remained in the model until 99 years old. Probabilities and state transition rates were obtained from a review of the literature. The rate of persistent symptoms or nerve root injury following ACDF was estimated at 3%, and the risk of postoperative cord injury was 0.1%.[6,15] The risk of postoperative hematoma requiring revision surgery was estimated at 0.001, and the risk of postoperative infection was estimated at 0.3%.[6,15] The rate of symptomatic pseudarthrosis after ACDF was estimated at 1.55%.[16] For cervical epidural injections, the risks of nerve root injury and cord injury were estimated at 0.25% and 0.21%, respectively.[11] The annual risk of ASD that requires surgical intervention after ACDF was estimated at 1.3%, whereas the annual risk of same or ASD after successful cervical epidural injections was estimated at 3% (which represents a similar risk of ASD of 1.3% with an added risk of subsequent same level disease).[17]

The described rate of successful management of cervical radiculopathy with epidural injections was not well described and significantly varied in the literature. Several in-depth systematic reviews have previously evaluated the topic and concluded that there are a sparse number of well-performed studies and reported heterogeneity across the studies.[9–12] Four studies that specifically evaluated avoidance of surgery with preoperative epidural injections were identified among the systematic reviews, with the rate of surgical avoidance ranging from 23% to 80%. Given this difficulty in establishing a point estimate, we performed our analysis across four different values for the success of injections.


Two identical cohorts of 10,000 patients were generated based on age-specific incidence of surgical cervical radiculopathy estimated from the 2012 National Inpatient Sample for International Classification of Diseases – 9- Clinical Modification code 81.02 and age adjusted based on incidence rates from an epidemiological study of cervical radiculopathy.[1,18] This population was used as an approximation of the annual rate of patients with persistent severe cervical radiculopathy failing 6 weeks of conservative management after the onset of symptoms with identifiable surgical pathology on advanced imaging.

Estimated Costs

Surgeon fees for single- or two-level ACDF and irrigation and débridement of deep neck abscess, practitioner fees for interlaminar cervical injections, and physical therapy fees for 6 weeks of therapy were obtained from the 2014 Medicare Physician Fee Schedule (Table 1 and Table 2). Facility fees for ACDF or revision procedures were obtained from 2013 Average Inpatient Medicare Payments; interlaminar outpatient facility fees were obtained from 2013 Average Outpatient Medicare Payments. Anesthesia fees were calculated using the 2013 Anesthesia National Conversion Factor × 10 base units + 4 time units for ACDF and × 3 base units + 4 time units for epidural injections. Respective provider and facility costs were summed for each procedure, and Medicare reimbursement rates were used for patients older than 65 years and converted to estimated private insurance reimbursement rates with a conversion factor of 1.39 based on a national survey of reimbursement rates.

Health Utilities

The quality-adjusted life year (QALY) value of an average individual in a well state was drawn from the Beaver Dam Health Outcomes study and has been previously used in cost-effectiveness analysis of cervical disk disease[19–21] (Table 3). The QALY associated with a state of radiculopathy was based on the health outcomes data presented in the ProDisc-C and BRYAN/PRESTIGE trials.[20–22] The same value for radiculopathy was used for the states of long-term dissatisfaction with surgery and nerve-root injury either after surgery or epidural injections. The QALY in the state of cord injury or paralysis was obtained from outcome surveys of spinal cord injury.[23,24] The disutility toll of primary ACDF or a series of cervical epidural injections was set at 0.1, the disutility toll of revision procedures set at 0.2, and the disutility toll of prolonged physical therapy was set at 0.1. The probabilities of adverse events (eg, nerve root injury, spinal cord injury, infection pseudarthrosis, ASD) that were used in the model were estimated based on the existing literature and can be found in Table 4.

Model Analysis

The model was simulated using a Markov chain Monte Carlo simulation with a cycle length of 1 year using TreeAge Pro (TreeAge Pro 2013; TreeAge Software). Costs were discounted 3% annually, and outcome measures were discounted 1.5% annually consistent with recent cost-effectiveness research recommendations.[22–24] A probabilistic sensitivity analysis (PSA) across a range of plausible values (all probabilities and values varied ±10%) with 10,000 iterations was performed to evaluate the robustness of the model.