DESIGN: MEDLINE, EMBASE, CINAHL were systematically searched (1990-April 2020) for studies describing the prevalence of NP and PS in knee and hip osteoarthritis using self-report questionnaires. Random-effects meta-analysis was performed. Statistical heterogeneity between studies and sub-groups (affected joint and population source as a proxy for disease severity) was assessed (I2 statistic and the Chi-squared test).
RESULTS: From 2,706 non-duplicated references, 39 studies were included (2011-2020). Thirty-six studies reported on knee pain and six on hip pain. For knee osteoarthritis, the pooled prevalence of NP was: using PainDETECT, possible NP(score ≥13) 40% (95%CI 32-48%); probable NP(score >18) 20% (95%CI 15-24%); using Self-Report Leeds Assessment of Neuropathic Symptoms and Signs, 32% (95%CI 26-38%); using Douleur Neuropathique (DN4) 41% (95% CI 24-59%). The prevalence of PS using Central Sensitization Inventory (CSI) was 36% (95% CI 12-59%). For hip osteoarthritis, the pooled prevalence of NP was: using PainDETECT, possible NP 29% (95%CI 22-37%%); probable NP 9% (95%CI 6-13%); using DN4 22% (95%CI 12-31%) in one study. The prevalence of possible NP pain was higher at the knee (40%) than the hip (29%) (difference 11% (95% CI 0-22%), P = 0.05).
CONCLUSIONS: Using self-report questionnaire tools, NP was more prevalent in knee than hip osteoarthritis. The prevalence of NP in knee and hip osteoarthritis were similar for each joint regardless of study population source or tool used. Whether defining NP using self-report questionnaires enables more effective targeted therapy in osteoarthritis requires investigation.