Psychometric evaluation and discriminative validity of the Short-Form Brief Pain Inventory across nociceptive,...

The measurement performance of the Short-Form Brief Pain Inventory (BPI-SF) across IASP-recognized pain mechanisms remains unclear. We evaluated the reliability and structural validity of the BPI-SF across clinician-assigned predominant nociceptive, neuropathic, and nociplastic chronic pain groups, and secondarily examined whether BPI-SF item patterns could distinguish these groups. In this cross-sectional cohort of 666 adults, 357 had predominant nociceptive pain, 125 neuropathic pain, and 184 nociplastic pain. Mean BPI-SF Severity and Interference scores were highest in the nociplastic group, followed by the neuropathic group and the nociceptive group, with significant pairwise differences across all groups. Internal consistency was excellent overall (Cronbach's α = 0.90, 95% CI 0.89-0.91) and within nociceptive (α = 0.91, 95% CI 0.89-0.92), neuropathic (α = 0.90, 95% CI 0.87-0.92), and nociplastic (α = 0.85, 95% CI 0.81-0.88) subgroups. Structural validity analyses supported a three-factor structure reflecting pain severity, emotional interference, and physical interference in the full sample and within all three mechanism subgroups. A confirmatory factor analysis model separating sleep as a fourth latent factor showed acceptable fit. Discriminative performance was limited: across five supervised machine learning models, the highest multiclass AUC was 0.70, and the strongest binary contrast involved nociplastic pain, with AUCs up to 0.75. Factor-informed subscale and reduced-item models did not improve discrimination. The BPI-SF is reliable and structurally valid across predominant pain mechanisms, but should not be used as a stand-alone mechanistic classification tool. PERSPECTIVE: This study supports the BPI-SF as a reliable multidimensional measure of pain severity and interference across predominant nociceptive, neuropathic, and nociplastic pain groups, while showing that BPI-SF item patterns alone are insufficient for mechanistic classification.