Background The Leeds Assessment of Neuropathic Symptoms (LANSS) and the painDETECT questionnaire (PDQ) are two validated screening tools for neuropathic pain (NP). Recent evidence reported a low level of agreement between these tests in knee Osteoarthritis. Several studies have recently applied the PDQ in Rheumatoid Arthritis (RA), suggesting a NP component in these patients, although the application and performance comparison with LANSS is yet to be studied.
Objectives Evaluate PDQ and LANSS performance for NP classification and investigate its optimal cutoff points in a RA cohort.
Methods Observational, cross-sectional study was designed including RA patients followed at our Rheumatology department. Patients with diagnosed neuropathy or non-RA risk factors for NP were excluded. Selected patients were evaluated in a medical visit where LANSS and PDQ were applied. Agreement between the two questionnaires was evaluated using kappa coefficient analysis. Receiver operating characteristic (ROC) analysis was performed using each tool as gold-standard and cutoff points to optimize agreement were investigated. Non-concordant patients were compared with concordant patients using parametric and non-parametric tests. Significance level was set as <0.05.
Results 112 RA patients were included, 86 (77%) were females, with a mean (SD) age of 55.1 (10.8) years and median disease duration of 13 years (range: 2–41). 102 (91%) were treated with DMARDs and 42% with a biologic DMARD. 45 (40%) patients had NP applying the LANSS (≥12) and 28% had NP in the PDQ (19 possible and 12 likely; no demographic or clinical significant differences were found between these two groups). 82 (73%) patients had concordant NP classification (59 negative, 23 positive) by the two tests. Concordant group had significantly superior median disease duration and inferior LANSS scores compared to non-concordant group (14 vs 12 years and 8 vs 13, respectively, p<0.05) with no other significant differences found. A moderate agreement (κ=0.41) and linear correlation (r=0.58, p<0.001) were observed between the two tests. In the ROC curve analysis, PDQ (≥13) showed an area under the curve (AUC) of 0.80, 95% CI [0.72–0.88] with a sensitivity and specificity of 51% and 88%, respectively, using LANSS as gold standard. LANSS (≥12) had an AUC of 0.80, 95% CI [0.71–0.90] and a sensitivity and specificity of 74% and 73%, respectively, using PDQ as gold standard. After ROC curve analysis, optimal cutoff for PDQ was 10, showing greater sensitivity (69%) but lower specificity (79%) with a slight increase in the agreement between the tests (κ=0.48). For the LANSS, the optimal cutoffs were the previous value or 13 (sensitivity 68% and specificity 78%) with a modest gain in the agreement (κ=0.42). Correction for both cutoff points resulted in a more substantial increase in agreement level (κ=0.51).
Conclusions In this study, LANSS and PDQ had a moderate level of agreement, possibly because they capture different dimensions of NP. New possible cutoffs were studied to increase agreement between the tests. Further studies with other conditions and a validated gold-standard for NP are needed to confirm this data.
Cohen J. Educ Psychol Meas. 1960;20:37–46.
Moreton B.J. et al Arthritis Care Res. 2015 Apr;67(4):519–28.
AW Christensen et al. Scand J Rheumatol 2016;1–9.
Disclosure of Interest T. Martins Rocha Grant/research support from: Portuguese Society of Rheumatology/Alfa Wassermann on May 2015, S. Pimenta: None declared, M. Bernardes: None declared, A. Bernardo: None declared, M. Barbosa: None declared, R. Lucas: None declared, L. Costa: None declared