Reply to the Comment by Finnerup and Gylfadottir Regarding ‘Lesion Without Pain, Pain Without Certainty: A Critical Reappraisal of the NeuPSIG Diagnostic Algorithm for Neuropathic Pain’ Authors: Didier Bouhassira, Nadine Attal Published: 2026 We thank the authors for their thoughtful comments (Finnerup and Gylfadottir 2026) and for engaging constructively with our proposal to reconsider the NeuPSIG grading system for neuropathic pain (Bouhassira and Attal 2026). We are pleased that they share our concerns regarding the limitations of the current framework, particularly the difficulty of inferring a causal relationship between a neurological lesion or disease and pain. This was one of the principal motivations for our proposal. The authors raise two important and complementary concerns regarding our convergence‐based model. First, a patient with a neurological lesion, sensory abnormalities and neuropathic pain descriptors could fulfil the proposed criteria despite pain arising from another mechanism, for example, osteoarthritis in a patient with diabetic polyneuropathy or spinal cord injury. Conversely, conditions widely accepted as neuropathic, including phantom limb pain, trigeminal neuralgia and inherited erythromelalgia, may lack sensory deficits or typical descriptors captured by neuropathic pain screening questionnaires and could therefore be insufficiently classified by our algorithm. We acknowledge these limitations. However, these examples also illustrate the inherent difficulty of designing a single framework encompassing the entire spectrum of neuropathic pain. Some, such as trigeminal neuralgia, represent specific clinical situations requiring management distinct from other neuropathic pain syndromes, while others, such as inherited erythromelalgia, are very rare. Such boundary cases should inform further refinement, but do not necessarily negate the framework's usefulness for the broader population evaluated for suspected neuropathic pain. The relevant question is not whether an algorithm can eliminate all misclassification, an unrealistic expectation in the absence of a diagnostic gold standard, but whether it can improve classification compared with the existing system while remaining applicable to routine practice. In this respect, explicitly incorporating neuropathic pain characteristics represents an important advantage. Screening questionnaires are imperfect and should never replace clinical judgement. Nevertheless, they have been extensively validated, with sensitivity and specificity generally in the 80%–90% range and diagnostic performance comparable to that of many tests conventionally regarded as more ‘objective’. Their inclusion adds standardized information directly related to the pain phenotype, a dimension largely absent from the current grading system. By contrast, the diagnostic performance of the existing NeuPSIG grading system as a whole has not been formally established. Incorporating validated descriptors may therefore reduce, although certainly not eliminate, both over‐ and under‐classification. This hypothesis should now be tested empirically. Simplicity is another potential advantage. The current NeuPSIG framework relies on sequential levels of evidence and ultimately gives a privileged role to investigations demonstrating a somatosensory lesion or disease. Our model instead relies on three readily assessable clinical dimensions: history and neuroanatomical plausibility, neuropathic pain characteristics, and sensory examination, without requiring complementary investigations to classify the nature of the pain. This simpler structure may facilitate implementation in routine, particularly where specialized investigations are unavailable. Importantly, these three dimensions should not be regarded as independent boxes whose arithmetic sum substitutes for clinical reasoning. Their diagnostic value depends on whether they form a coherent clinical picture. In a patient with osteoarthritis and a coexisting neurological disorder, burning pain, sensory loss and a neurological lesion may coexist without reflecting the same mechanism. Conversely, phantom limb pain, trigeminal neuralgia and inherited erythromelalgia illustrate that neither typical neuropathic descriptors nor demonstrable sensory loss is universally present. Such exceptions highlight the limitations of any rigid additive score and the continuing need for clinical judgement. We therefore agree that no grading system can eliminate clinical judgement. However, we would be cautious about restoring ‘objective confirmation of a neurological lesion’ as an alternative criterion for probable neuropathic pain. This risks reintroducing the conceptual problem we sought to address: treating evidence for a lesion as evidence for the nature of the pain. Complementary investigations remain essential for identifying neurological disease and clarifying aetiology, but their meaning with respect to pain necessarily depends on the clinical context. The suggestion to retain only ‘possible’ and ‘probable’ categories nevertheless deserves consideration. ‘Definite neuropathic pain’ may imply a degree of certainty that cannot generally be achieved when no test can demonstrate that a particular lesion is causally responsible for an individual's pain. Our terminology retained the familiar NeuPSIG categories, but the authors' comments underscore the importance of distinguishing increasing clinical confidence from proof of causality. Our proposed algorithm should therefore be regarded as a conceptual framework and starting point for discussion rather than a finalized diagnostic instrument. Its performance, including the false‐positive and false‐negative classifications highlighted by the authors, requires prospective evaluation. Nevertheless, its potential advantages over the current system include standardized pain phenotyping using validated tools, avoidance of an intrinsic hierarchy between patient‐reported and lesion‐based evidence, and a simpler framework potentially more applicable to clinical practice. We thank the authors for initiating this discussion and welcome their suggestion that NeuPSIG consider convening a working group to examine potential revisions to the grading system. Conflicts of Interest The authors declare no conflicts of interest. Linked Articles This article is linked with Finnerup and Gylfadottir (2026) and Bouhassira and Attal (2026). To view these papers, visit https://doi.org/10.1002/ejp.70381 and https://doi.org/10.1002/ejp.70350. References