Background/Purpose: Ongoing pain state in rheumatoid arthritis (RA) often persists after the resolution of inflammation, indicating the transition between the acute inflammatory pain and post-inflammatory persistent pain phenotypes. PainDETECT (PDQ) was developed as a self-reported neuropathic pain screening tool. We conducted a Rasch analysis to investigate whether measurement properties of PDQ can be used as an outcome measure.

 Methods: For RA subjects enrolled in the University of Pittsburgh Rheumatoid Arthritis Comparative Effectiveness Registry (RACER), Rasch analysis was conducted for all RACER patients who completed PDQ and PROMIS29 short form. Unidimensionality, reliability, item difficulty, category functioning, and differential item function were examined using the partial credit model for polytomous items (WINSTEPS ver. 3.93.2). Differential item function (DIF) analyses for gender and age groups (<60,60-70,>70) were performed to examine item invariance for different groups.

Results:   For the 302 subjects analyzed, age was 63.8 ¡¾ 12.4 (mean ¡¾ SD) years with disease duration of 18.4¡¾11.9 years. Given misfit and high fit residuals, time course and radiating pain items were removed. Remaining 7-item PDQ fit the Rasch model (Table 1). Item residuals showed high correlation between the burning and slight pressure items and these items were treated as a testlet. Cold-or-heat item exhibited marginally disordered threshold but the rescoring did not substantially affect fit (Table 2). DIF analyses for gender and age groups showed uniform DIF. Person-item distribution showed that PDQ was reasonably targeted (Figure).

Conclusion:   Rasch analysis of 7-item PDQ suggests that PDQ may function as an outcome measure and may provide a useful tool to predict RA treatment outcome for neuropathic pain.