Is the Neck Disability Index an Appropriate Measure for Changes in Physical Function After Surgery for Cervical Spondylotic Myelopathy?

INTRODUCTION

Patient-reported outcome measures (PROs) are tools that allow clinicians and researchers to assess patient-reported health status for physical, mental, and social well-being.¹ With the passing of the affordable care act in 2010, there has been a shift from volume-based to value-based care.² As a result, it is becoming increasingly important that the most apt PRO measures are identified for the appropriate patient populations.

The Neck Disability Index (NDI), a PRO measure first developed by Vernon in 1991, is a self-assessment measure used to assess disability in patients with neck pain.³ It was developed using the Oswestry Disability Index, a PRO measure used in patients with low back pain, as a model.⁴ Initial validation for the NDI considered only “whiplash” injury patients treated in an outpatient setting; however, the NDI is now widely used to evaluate the efficacy of cervical surgery.^5–9 The NDI consists of 10 domains—pain intensity, personal care, lifting, reading, headache, concentration, work, driving, sleep, and recreation—designed to assess the level of disability in patients with neck pain. Patients self-report their level of function/disability in each domain, after which a composite score is calculated.

As the first PRO measure specific to patients with neck pain, the NDI is arguably the most widely used PRO for patients with neck disorders.¹⁰ The use of scales, such as the NDI, presents a unique challenge when evaluating patients undergoing treatment for cervical spondylotic myelopathy (CSM) where function, not pain, is the primary symptom. The NDI is still commonly used for these patients despite a lack of validation in this particular population.^6–9 The Short Form-36 (SF-36) and the Short Form-12 (SF-12) are 2 PRO measures that provide a global assessment of health, have strong internal psychometric properties, and have been used and validated in a wide variety of patient populations, including those experiencing CSM.^11–13 SF-36 has previously been used in studies as an anchor for analysis when looking at the NDI and other PRO measures; therefore, in this study we used the abbreviated SF-12 as an anchor in the current analysis.^14,15

The purpose of this study was to determine whether the NDI is a valid patient-reported outcome measure for tracking changes in physical function after surgical intervention for cervical spondylotic myelopathy.

PATIENTS AND METHODS

RESULTS

DISCUSSION

There are conflicting data on the performance and applicability of the NDI in patients undergoing surgical or nonsurgical treatment for cervical spondylotic myelopathy.¹⁰ CSM, the most common degenerative condition of the spine, tends to present with muscle weakness, gait imbalance, and urogenital dysfunction, among other symptoms.¹⁶ It is a progressive disorder, and although surgery is designed to prevent the progression of the disease, it may not lead to complete resolution of symptoms. Based on this trend, surgery is often preformed when there is compression on the spinal cord, but the symptoms are mild. Furthermore, when patients present late and have severe, longstanding myelopathic symptoms, surgery is unlikely to result in a dramatic improvement of neurologic function. Although the SF-36 and the abbreviated SF-12 show excellent sensitivity to changes in physical and mental health status in patients undergoing surgery for CSM, no validation studies have been done for NDI in patients with CSM despite its widespread use.^13,22 As health care shifts from volume-based to value-based reimbursements, it is critical that appropriate metrics be used to establish the value of care.

The present analysis found all domains of the NDI, NDI composite, and PCS scores to significantly improve after surgery. Improvements in the work and recreation domains, as well as the composite NDI score, were found to be the only significant predictors of change in PCS after surgery for CSM. Even though the NDI pain component was not a significant predictor of PCS, on average this domain displayed significant improvement—a finding consistent with prior literature.²³ Most other subdomains followed the pattern of the pain domain by demonstrating improvement but not being significant predictors for (Δ)PCS in the linear regression model. The present study's outcome data are comparable to what has been reported in the literature with respect to NDI and PCS scores. Several studies have reported baseline NDI and PCS scores that were similar to the current study's preoperative values, reporting significant improvement in scores from baseline to 1 to 2 years postoperatively in both NDI and PCS components.^9,15,23

Existing literature is relatively sparse regarding improvement in work and recreation domains for surgically treated CSM patients. However, multiple studies have found that upper limb function has better recovery after surgery, followed by lower limb function.^24–26 These improvements in limb function directly correlate with the SF-12 PCS score and may also explain why the work and recreation domains have stronger associations with PCS in this study. Here, the work and recreation domains made up 2 of 4 of the highest scores measured at baseline, and these questions—of all questions on the NDI—seem to be most similar to those asked in the SF-12 survey.

If NDI is to be used as an outcome measure in the surgically treated CSM population, then a better than moderate association with PCS in all or most all domains found on NDI is expected. Instead, only 2 domains, work and recreation, were significant predictors of improvement in physical function over time. This suggests that although other domains of the NDI may demonstrate statistical improvements after surgery, they do not necessarily translate to physical function changes. This is possibly related to the fact that patients with CSM exhibit signs of gait impairment—a symptom that is not addressed in any components of the NDI.¹⁶

This is one of the first studies to directly study the individual NDI domains instead of the aggregate NDI score. There have been several studies questioning the reliability, validity, performance, and applicability of the composite NDI score in patients with cervical radiculopathy and myelopathy.^10,27 Cleland et al²⁷ found the NDI to have adequate test-retest reliability, but poor construct validity in a population of patients with cervical radiculopathy. The authors found that the NDI was not an effective tool to identify changes in patients' perceived levels of disability when such a change had occurred.²⁷ In another study, Hung et al¹⁰ found the NDI to have a high level of unexplained variance, which the authors interpreted as “poor unidimensionality”—considering that the NDI is supposed to measure one condition, a lower percentage of unexplained variance was expected. Additionally, it has been suggested that the presence of psychologic distress may have confounding effects on NDI scores.²⁸

If NDI proves to be ineffective in tracking physical function changes in patients receiving surgical intervention to address CSM, there needs to be a set of standardized outcome measures to appropriately assess this particular parameter. Kalsi-Ryan et al²² claimed that because CSM has such a heterogenous presentation, a single outcome measure is not sufficient to quantify the broad range of neurologic deficits seen in this population. The authors identified additional outcome measures that characterize the deficits in the CSM population with greater validity, reliability, and responsiveness—including QuickDASH, Berg balance scale, GRASSP, grip strength, GAITRite analysis, and the 30-meter walk test—suggesting that with the use of these scales, clinicians would better be able to assess outcomes and monitor the natural course of the disease.²² Recently, there has been the development of a new type of outcome measure—Patient Reported Outcomes Measurement Information System (PROMIS)—which is being used in the field of orthopaedics.¹¹ What distinguishes PROMIS outcomes from the traditional PRO measures is use of item response theory.¹¹ Item response theory uses computerized adaptive testing, which after starting at a predefined point on the scale, chooses each subsequent item based on answers to previous items.¹¹ Studies have shown that the psychometric characteristics of the PROMIS questionnaire compared favorably with a traditional outcome measure for the upper extremity.²⁹ Patient time burden was also significantly reduced when using the PROMIS questionnaire compared with the traditional questionnaire.²⁹ Investigation into how these instruments preform in the spine population is a potential area for future exploration—one that should serve to further elucidate the most appropriate outcome assessment tool in the investigation of physical function changes over time for patients undergoing surgery to prevent the progression of CSM.

Limitations to this study include the fact that data were retrospectively reviewed and that only the PCS portion of the SF-12 survey was used as an anchor. In order to have a more robust analysis, the use of additional anchors—including patient satisfaction with the surgery—would have been beneficial. Additionally, CSM patients often have a psychological component to their disability, so the use of the PCS as an anchor could possibly be confounded by not including a separate analysis of the MCS portion of the SF-12.

CONCLUSION

In summary, most domains of the NDI—including the pain subdomain—did not significantly predict PCS change after surgery for myelopathy. It is important for an outcome measure to consider all aspects of impairment in the CSM population and the full impact on functionality if it is to be used in this set of patients. Based upon these results, the individual item bank and composite scoring of the NDI are inappropriate for evaluating quality of life and physical function changes in patients who are surgically treated to prevent the progressive symptomatology of CSM.