Sensitivity to Change and Internal Consistency of the Northwick Park Neck Pain Questionnaire and Derivation of a Minimal Clinically Important Difference

Abstract

To estimate the sensitivity to change and internal consistency of the Northwick Park Neck Pain Questionnaire (NPQ) and derive its minimal clinically important difference (MCID). Data on 311 participants in a randomized controlled trial of physical therapy for neck pain were analyzed, between baseline and 6-month follow-up. Cronbach alpha was used to estimate internal consistency. Three methods of evaluating the NPQ's sensitivity to change were used; Cohen's effect size, standardized response mean, and Guyatt's responsiveness statistic. The smallest change on the NPQ that can be construed as clinically important (the MCID) was obtained by (1) comparing values derived from different methods that account for the measurement error of the questionnaire with improvements in other neck pain-specific measures, and then (2) examining the impact of the addition of an independent measure ("anchor") relating to participants' subjective rating of improvement. Cronbach alpha was 0.79 for the whole scale. Using all participants, Cohen's d was 0.78, standardized response mean was 0.71, and Guyatt's responsiveness statistic was 0.93. Using only participants who stated they had improved, the estimates were 1.35, 1.27, 1.45, respectively. The MCID for the NPQ was determined as a 25% reduction in score from baseline together with a patient's global rating of their neck pain of at least "better." These results show high internal consistency and sensitivity to change for the NPQ, and provide an MCID that allows participants with varying levels of severity to demonstrate improvement.