Development and Validation of the Pain Impact Survey

The aim of this study was to develop and validate a brief self-report tool to assess chronic pain impact in people with chronic diseases and the general population. Sixty-five pain items were drawn from an item pool that contained items from 10 of the most commonly used pain measures. The items were administered through the internet to a convenience sample of 782 people. On the basis of statistical item analysis using classical and item response theory (IRT) methods, 8 items were selected because they had the best item characteristics and spanned the desired content of severity and impact. In a following validation study this 8-item pain impact survey was then administered by the internet to a nationally representative sample of the general population stratified by age, gender and income. Preliminary analyses on 829 general population participants revealed that the Pain Impact Survey is unidimensional (CFI=0.95, TLI=0.92). There was no indication of differential item functioning with regard to gender, race, ethnicity, level of education, religion, spiritual practice, and medical condition, suggesting that the items function similarly across these subgroups. IRT-based scoring of pain impact provided enhanced discrimination across all levels of pain impact, particularly at very low and very high levels (i.e., reducing floor and ceiling effects). Construct validity was supported by strong correlations with numerical rating and visual analogue scale items that are commonly used to rate pain severity in hospitals (r=0.81–0.84, p<0.01). Discriminant validity was suggested by differences in reported pain impact by medical condition, with lowest levels reported by people with no medical condition and highest scores reported by people with constipation or lung disease. We conclude that the Pain Impact Survey is an efficient and precise assessment tool.