Validation of Neck Disability Index Severity among Patients Receiving One or Two-Level Anterior Cervical Surgery

Article information

Abstract

Introduction

Neck-related pain and disability have a reported lifetime incidence of 67% in the general population [1]. When conservative management proves ineffective, anterior cervical discectomy and fusion (ACDF) or cervical disc arthroplasty (CDA) is used to treat cervical spine degeneration, which is often responsible for symptoms. Many clinicians opt to focus on patient-reported outcome measures (PROMs), which provide an understanding of symptoms from the patient’s perspective, to accurately assess recovery from ACDF and CDA. The Neck Disability Index (NDI) is one such PROM that measures the level of disability due to pain among cervical spine patients using a set of 10 different questions [2,3].

Previous studies have verified the NDI questionnaire against other PROMs in ACDF and CDA populations [4,5]. While the NDI score is well validated as an accurate measure of patient symptoms, the literature is less clear about the context or relative severity of any given score. Several previous studies have categorized disability severity using the NDI. Vernon and Mior [6] assessed the reliability of NDI in whiplash patients and first stratified NDI severity as none (0–4), minimal (5–14), moderate (15–24), severe (25–34), and complete (≥35). Other studies on whiplash injury and ACDF patients have also divided baseline NDI scores according to severity; however, the threshold scores used in each study varied greatly [7–9] and the selection criteria for NDI score thresholds and their corresponding severity groups are not well documented.

To our knowledge, no studies have validated the accuracy and reliability of NDI severity categories in ACDF or CDA patients [10]. While previous studies have categorized patients, a single stratification method has yet to be statistically confirmed and established as more than arbitrary groupings of scores in those with degenerative disease of the cervical spine. Given that neck disability severity may play an important role in overall postoperative recovery, validation of specific categories may provide better context for NDI use during preoperative planning and for identification of symptoms that require greater focus in the postoperative period. Therefore, the present study aimed to evaluate the validity of NDI severity categories against various other PROMs among patients undergoing ACDF or CDA procedures.

Materials and Methods

Results

A total of 290 patients met the inclusion/exclusion criteria. The patients had a mean age of 48.7 years, mean BMI of 29.5 kg/m², and 60.0% were male. Most patients were of Caucasian ethnicity (77.5%), nonsmokers (85.9%), and had an ASA classification of ≥2 (79.1%). The mean CCI was 0.62 and the most common medical comorbidities were hypertension (23.1%), arthritis (9.7%), and diabetes mellitus (8.3%). None of the patients had a diagnosis of AIDS, neurological disease, paraplegia, peripheral vascular disease, or metastatic cancer (Table 1). Most patients had a preoperative spinal pathology of herniated nucleus pulposus (81.7%), experienced symptoms of myeloradiculopathy (85.2%), and had a single level surgery (64.1%). The mean operative duration was 57.9 minutes, mean EBL was 30.3 mL, and mean postoperative length of stay was 11.5 hours. Most patients were discharged on POD 0 (Table 2).

Table 1

Patient demographics (N=290)

Table 2

Perioperative characteristics (N=290)

NDI, VAS neck, VAS arm, and SF-12 PCS were significantly improved at all postoperative timepoints over 2 years (p<0.001, all). SF-12 MCS significantly improved over 1 year (p≤0.039, all). VAS neck and SF-12 PCS demonstrated statistically significant, strong correlations with NDI at all timepoints (p<0.001, all). VAS arm demonstrated strong correlations with NDI preoperatively and at 12 weeks, 6 months, and 2 years postoperatively and showed moderate correlations at 6 weeks and 1 year (p<0.001, all). SF-12 MCS demonstrated strong correlations with NDI from 6 weeks to 2 years and moderate correlations at the preoperative timepoint (p<0.001, all) (Table 3). The mean scores for VAS neck, VAS arm, SF-12 PCS, and SF-12 MCS differed significantly among NDI severity groups at all timepoints (p≤0.009, all). No patients were categorized as having “complete” disability at 2 years (Table 4). Between-group comparisons of NDI categories revealed no significant differences in VAS neck scores between the “none” and “mild” groups preoperatively (1.5±0.70) and 2 years postoperatively (1.4±0.76), “moderative” and “severe” groups at 2 years (1.6±0.86), “moderate” and “complete” groups at 2 years (1.6±0.86), and “complete” and “severe” groups at all timepoints. VAS arm scores showed no significant differences between the “none” and “mild” groups preoperatively and at 6 weeks, 12 weeks, and 2 years, “none” and “moderate” groups at 6 weeks, “mild” and “moderate” groups at 6 weeks, 1 year, and 2 years, “mild” and “severe” groups at 1 year, “mild” and “complete” group at 1 year, “moderate” and “severe” groups at 12 weeks to 1 year, “moderate” and “complete” groups at 1 year, and “severe” and “complete” groups at all timepoints. VAS neck and VAS arm scores showed significant differences for all other between-group comparisons (Table 5). SF-12 PCS scores showed no significant differences between the “none” and “mild” groups at 2 years, “mild” and “moderate” groups at 2 years, “mild” and “severe” groups at 6 months, “mild” and “complete” groups at 6 months, “moderate” and “severe” groups from 6 weeks to 2 years, “moderate” and “complete” groups at 6 weeks to 2 years, and “severe” and “complete” groups at all timepoints. SF-12 MCS scores showed no significant differences between the “none” and “mild” groups at any timepoint, “none” and “moderate” groups preoperatively and at 2 years, “mild” and “moderate” groups preoperatively and at 2 years, “moderate” and “severe” groups at 6 months and 2 years, “moderate” and “complete” groups at 1 year, and “severe” and “complete” group at all timepoints. SF-12 PCS and SF-12 MCS scores significantly differed between NDI severity groups at all timepoints (Table 6).

Table 3

Patient-reported outcomes

Table 4

Mean PROM scores by NDI severity

Table 5

Discriminant validity of NDI vs. VAS

Table 6

Discriminant validity of NDI vs. SF-12

Discussion

ACDF and CDA are established treatment options that show favorable results for the treatment of various cervical spine pathologies [11]. The increased use of PROMs to assess patients’ perception of their health has led to NDI becoming the most widely used tool to measure an individual’s associated neck disability [3]. Although NDI is a strongly validated measurement in the spine, the severity thresholds used to stratify patients based on their disability are not well established. Therefore, the present study evaluated the validity of previously established NDI severity thresholds against other well-established PROMs in patients undergoing ACDF or CDA.

NDI has an established correlation with various PROMs, making it a useful tool for understanding patients’ disabilities and how they may affect other aspects of their health and quality of life. The present study showed that NDI demonstrated a moderate-to-strong correlation at most timepoints for VAS neck, VAS arm, SF-12 PCS, and SF-12 MCS. Steinhaus et al. reported a significant correlation between NDI and both Patient-Recorded Outcomes Measurement Information System physical function and SF-36 PCS in adults undergoing cervical spinal surgery [12]. Additional studies have shown NDI to be correlated with VAS, SF-36 MCS, and Hospital Anxiety and Depression Scale among a variety of other PROMs [3,13]. The present study provided further context for the NDI questionnaire by investigating the validity of the severity thresholds to understand both the limitations and capabilities of NDI as a meaningful clinical tool.

The use of NDI severity thresholds may help to quantify patients’ perceived disability by providing additional context for scores. There are several proposed scoring breakdowns in the published literature [10] and the present study used the thresholds established by Vernon and Mior [6], the original creators of the NDI tool, as follows: none (0–4); minimal (5–14); moderate (15–24); severe (25–34); and complete (≥35). However, the suggested thresholds have yet to be validated against other PROMs used in the cervical spine. McCarthy et al. [14] used a similar scoring system to validate NDI against SF-36 and reported percentage scores of 0%–20% (normal), 21%–40% (mild disability), 41%–60% (moderate), 61%–80% severe, and ≥80 (complete). In the present study, the NDI severity categories showed significant differences in mean VAS neck, VAS arm, SF-12 PCS, and SF-12 MCS at all timepoints. Furthermore, individuals categorized as complete disability reported higher VAS pain scores and lower mental health and physical function scores than those in the none and mild categories. This suggests that the threshold values originally proposed by Vernon and Mior [6] are not completely arbitrary and may help to improve preoperative education and understanding of postoperative recovery in patients.

The direct assessment of discriminant validity in the present study enabled a more direct comparison between different levels of severity and an understanding of whether these scoring thresholds could be used to correctly distinguish between patients with different levels of disability. The results of the present study demonstrate that scores are not as well differentiated at the extremes of the severity spectrum (i.e., none versus mild, severe versus complete). It is important to note that the lack of differentiation at the extremes may be partly due to the patient population included in the present study. Patients undergoing procedures for degenerative cervical pathologies tend to experience at least some disability and it is likely that this population did not include a significant number of patients with severe spinal cord damage.

Our results revealed that differences between NDI severity groups were not uniform for all PROMs. Assessment of discriminant validity yielded much more consistent results for VAS neck and arm than SF-12 MCS or SF-12 PCS, which could be due to the nature of the questionnaires. While SF-12 is a general quality of life measurement intended for use in a variety of patient populations, VAS may capture symptoms related to the individual’s cervical pathology more effectively by only focusing on questions related to neck and arm pain. It is important to note the differences between SF-12 MCS and SF-12 PCS. While SF-12 PCS demonstrated significant results at all but one preoperative timepoint, SF-12 MCS showed no significant results between the none and mild, none and moderate, mild and moderate, and complete and severe groups preoperatively. This may be due to a stronger relationship between disability and pain or physical function compared with mental health. These findings are supported by those reported Stull et al. [15], who found that individuals with significant preoperative neck pain experienced significant improvement in neck pain, disability, and physical function following ACDF, whereas only a subset of their patient population experienced significant improvements in mental health, as measured by SF-12 MCS.

A comparison of the discriminant validity among the different PROMs showed that VAS neck yielded the most consistent differences among severity groups, even compared with VAS arm. This suggests that NDI may be a better representative of neck pain than arm pain, which is an important relationship to acknowledge when using the metric to assess a patient’s symptomatology. Narain et al. [16] examined a cohort of patients undergoing ACDF and showed minimum clinically important difference achievement rates for VAS neck, VAS arm, and NDI of 55.4%, 36.9%, and 76.6%, respectively. This further highlights that while patients may recover well in terms of their neck pain and disability, it is more difficult for them to achieve clinically significant improvement in arm pain.

The present study revealed that PROM scores are not as well differentiated among severity groups at later time points. For example, there were no significant differences between the mild and complete groups at 6 months for SF-12 PCS or at 1 year for VAS arm. Additionally, while VAS neck demonstrated significant differences at all previous timepoints between the complete and moderate groups, there was not significant differentiation at 1 year. However, this may be because patients recover over time after their procedure. Andresen et al. found that patients undergoing ACDF demonstrated significant improvement in disability following surgery, with a mean preoperative NDI score of 40.0, which improved to 22.7 in the postoperative period [17]. Although disability likely improves postoperatively, discriminant analysis would still be expected to yield significant results if pain, physical function, and mental health also improved. However, it is possible that although a patient’s disability may drastically improve following surgery, there may be additional factors unrelated to disability that contribute to their continued pain and limited physical function or poor mental health. This suggests that NDI may be a useful tool for preoperative stratification, but as patients progress in their postoperative recovery the metric may become less useful for differentiating severity of disability. Additionally, the extent of recovery following surgery may limit the number of individuals experiencing severe or complete disability at later timepoints, making any comparison among differing severity groups more difficult.

The present study has some limitations that should be considered. First, the use of PROMs may introduce response or recall bias, as occurs with all self-reported questionnaires used to assess outcomes. The study included outcomes up to 2 years; therefore, it was limited by the number of patients that completed the questionnaires throughout the entire follow-up period. Furthermore, the use of a patient cohort that underwent surgery performed by a single surgeon at a single institution, as well as the limited number of individuals at the extremes of disability, may limit the generalizability of our results to other populations. Additional studies that include a multicentered, multiple surgeon cohort will strengthen the results of the present study. Finally, the included patient cohort specifically underwent anterior cervical operation at one or two levels; therefore, our findings may not be generalizable to individuals undergoing surgery at three or more levels, myelopathy patients receiving posterior cervical operation, particular revision procedures, and those receiving surgery for deformity, malignancy, trauma, or infectious etiology.

Conclusions

Neck disability, as measured by NDI, is strongly correlated with neck and arm pain (VAS), physical function (SF-12 PCS), and mental health (SF-12 MCS) and demonstrates generally worse outcomes with increasing severity. However, differences between groups were not uniform for all PROMs. Specifically, severity categories were not as applicable at extremes or longer postoperative timepoints and demonstrated more consistent results for VAS than SF-12. This suggests that NDI and its established severity thresholds may be more useful as a preoperative stratification tool than a way to assess disability during postoperative recovery. Further studies are required to develop a more clinically relevant severity threshold for patients undergoing cervical spinal surgery.

Article information Continued