Background 10-20% of patients with rheumatoid arthritis (RA) display features of central sensitization. The numbers are less well documented within psoriatic arthritis (PsA) and other spondyloarthritis (SpA). There is a need for instruments that can be used to identify underlying pain mechanism. The painDETECT Questionnaire (PDQ) was originally developed and validated to classify pain into unlikely (<13 points), unclear (13-18 points) or likely neuropathic pain (>18 points). Current evidence suggests that the PDQ may also assist identification of pain that is primarily driven by central mechanisms.
Objectives To evaluate the psychometric properties of the painDETECT Questionnaire (PDQ) when used to classify pain in inflammatory arthritis by applying Rasch analysis and to explore the reliability of pain classification.
Methods For the Rasch analysis 306 (102x3) questionnaires from patients with RA, PsA or other SpA were extracted from “the DANBIO painDETECT study” (cross sectional registry survey N=7918, not published). The analysis was directed at the seven items assessing somatosensory symptoms. Three points were addressed: 1) the performance of the six-category Likert scale (Partial Credit Model); 2) whether a unidimensional construct was defined; and 3) the reliability and precision of estimates. The Rasch computer program WINSTEPS 3.68.2 was used.
Another group of 30 patients diagnosed with RA, PsA or other SpA participated in a test-retest study. Re-test took place within two-four days. Intraclass Correlation Coefficients (ICC) and classification consistency were calculated.
Results Rasch analysis: (1) acceptable psychometric properties of the rating scale; the frequency distribution peaked in category 0 except for item 5. Threshold calibration >10 observations per category except for category 5 in items 2, 3 and 4. No disorder in the category measures for all items. Scale category outfit; Mnsq <2.0. (2) The standardized residuals showed that 55.4% of the total variance was explained by the measure and the Eigenvalue was 1.6 both supporting unidimensionality. (3) No misfitting items. (4) Different item hierarchies across diagnoses. (5) A reasonably targeted item-person map and person and item separation indexes of 2 (reliability=0.80), and 7.35 (reliability=0.98).
Test-retest: ICC: RA 0.86 (0.56-0.96), PsA 0.96 (0.74-0.99), SpA 0.93 (0.76-98), in total 0.94 (0.84-0.98). Classification consistency was: RA 70%, PsA 80%, SpA 90%, in total 80%.
Conclusions The results support the use of the PDQ in future research of mechanism-based pain classification in inflammatory arthritis. In clinical care, the PDQ may serve as an easily applied screening tool assisting identification of individuals with pain suspected not to originate from peripheral mechanisms of nociception and inflammation.
Freynhagen R et al. painDETECT: a new screening questionnaire to identify neuropathic components in patients with back pain. Curr Med Res Opin 2006 Oct;22(10):1911-20.
Koroschetz J et al. Fibromyalgia and neuropathic pain-differences and similarities. A comparison of 3057 patients with diabetic painful neuropathy and fibromyalgia. BMC Neurol 2011;11:55
Acknowledgements The authors wish to thank all participating Departments of Rheumatology and DANBIO. Furthermore, The Oak Foundation and Selsbjerg Holding for financial support.
Disclosure of Interest None declared