Measurement properties of the neck disability index
From: J Orthop Sports Phys Ther. 2009 May;39(5):400-17
Patient completed questionnaires are commonly used to measure clinical outcome in efficacy studies. Outcome questionnaires for neck pain should measure the effect of treatment on pain, motion, disability, activities of daily living, social function, and work function. The Neck Disability Index is a commonly used neck pain questionnaire. It is modeled after the Oswestry Back Disability questionnaire and includes 10 self-report items covering activities of daily living (7 items), concentration (1 item), and pain (2 items). Responses are on a 0-to-5 Likert scale with total score ranging from 0 (no pain or disability) to 50 (severe pain and disability). It has been studied in patients with whiplash associated symptoms and in groups with mixed causes for neck pain. It has excellent test-retest properties and good convergent validity when compared with the McGill Pain Questionnaire and a global improvement scale. Item analysis suggested that the Neck Disability Index measures a single domain.
Systematic review of clinical measurement to find and synthesize evidence on the psychometric properties and usefulness of the neck disability index. The neck disability index is the most commonly used outcome measure for neck pain, and a synthesis of knowledge should provide a deeper understanding of its use and limitations. Using a standard search strategy (1966 to September 2008) and 4 databases (Medline, CINAHL, Embase, and PsychInfo), a structured search was conducted and supplemented by web and hand searching. In total, 37 published primary studies, 3 reviews, and 1 in-press paper were analyzed. Pairs of raters conducted data extraction and critical appraisal using structured tools. Ranking of quality and descriptive synthesis were performed.
Horizon estimation suggested the potential for 1 missed paper. The agreement between raters on quality assessments was high(kappa = 0.82). Half of the studies reached a quality level greater than 70%. Failures to report clear psychometric objectives/hypotheses or to rationalize the sample size were the most common design flaws. Studies often focused on less clinically applicable properties, like construct validity or group reliability, than transferable data, like known group differences or absolute reliability (standard error of measurement or minimum detectable change. Most studies suggest that the neck disability index has acceptable reliability, although intraclass correlation coefficients range from 0.50 to 0.98. Longer test intervals and the definition of stable can influence reliability estimates. A number of high-quality published (Korean, Dutch, Spanish, French, Brazilian Portuguese) and commercially supported translations are available. The neck disability index is considered a 1-dimensional measure that can be interpreted as an interval scale. Some studies question these assumptions. The minimum detectable change is around 5/50 for uncomplicated neck pain and up to 10/50 for cervical radiculopathy. The reported clinically important difference is inconsistent across different studies ranging from 5/50 to 19/50. The neck disability index is strongly correlated to a number of similar indices and moderately related to both physical and mental aspects of general health.
The neck disability index has sufficient support and usefulness to retain its current status as the most commonly used self report measure for neck pain. More studies of clinically important difference in different clinical populations and the relationship to subjective/work/function categories are required.
