Abstract
The Neck Disability Index (NDI) is the most commonly used patient reported outcome (PRO) for cervical assessment in clinical and research settings. It is strongly advocated by professional associations and statutory body guidelines. However, two 2009 published studies respectively used 'Maximum Likelihood Extraction'-Factor Analysis (n = 61) and 'Rasch Methodology' analysis (n = 521) that demonstrated the NDI is unsound. This was due to the bidimensional structure that results in ordinal rather than interval scaling and questions the fundamental practice of calculating change scores and other parametric statistics on NDI data. To test these criticisms and analyse overall NDI performance, primary care data from a convenience sample (n = 95, responses = 284, invalid = 44) was analysed. Patients' conditions expressed diversity in region and severity that included non-specific, soft-tissue, joint, whiplash and traumatic injuries. Overall PRO performance was assessed using the Measurement of Outcomes Measures (MOM) quantitative scale (25 × 3-point items = 100%) and the 'Bot' quantitative scale (12 dichotomous items). Psychometric properties were calculated and included factor structure, whilst practical characteristics included completion and scoring time and missing responses. In addition the total and individual item with-in tool scores were assessed and distribution analysed. The NDI was found to have suboptimal overall performance on both the MOM and Bot scales (66%) due to poor readability, inadequate distribution, narrow within-tool range, bi-dimensional factor structure and high scoring time in the presence of missing responses (8%). The study supports recent criticism and recommends the NDI not be used as a clinical or research PRO for cervical assessment. The importance of this work: patient reported outcome (PRO) measures for use in the clinical and research setting require a solid scientific support basis for both their methodological and practical characteristics. Much of the current research on cervical injuries and whiplash uses the NDI. However, in the recent literature, including that from respected Australian research and academic institutions, the properties of the tool itself are rarely if ever calibrated by the researchers themselves. This study supports the recent publications that have analysed the NDI and found that the recognised reasons for recommend use are predominantly subjectively based or rely upon psychometric properties and analysis performed up to 2 decades previously. There is a need to calibrate an instrument prior to its use in a research study or use standards from a study that has investigated either the same patient pool or alternatively a broad and diverse population that will have lower or more conservative values. This paper contributes to these areas of debate by highlighting recent concerns within the frame work of broader analysis and negative criticisms of the NDI.