Objective: To compare measurement properties of the P4 pain scale, Western Ontario and McMaster Universities Arthritis Index pain subscale (WOMAC-pain), and Intermittent and Constant Osteoarthritis Pain (ICOAP) measure in patients with knee osteoarthritis (OA).

Design: A secondary analysis from a randomized controlled trial included participants (n = 156) with knee OA that were consulting with a surgeon regarding knee arthroplasty. They completed pain measures (P4, WOMAC-pain, ICOAP) and WOMAC-function subscale (WOMAC-function) at baseline and 2 weeks. Measurement properties assessed in various subgroups included floor/ceiling effects, test-retest reliability using intraclass correlation coefficients (ICC2,1), internal consistency using Cronbach's ɑ, factorial structure of each pain measure combined with WOMAC-function using principal component analysis, and responsiveness using standardized response mean (SRM).

Results: P4 had low floor and ceiling effects (<1%). P4 test-retest reliability (ICC2,1 = 0.72), internal consistency (Chronbach's ɑ = 0.91), and responsiveness (SRM = 0.56) were similar to the values for WOMAC-pain and ICOAP. Factorial structure of P4 and ICOAP were separate from WOMAC-function items. WOMAC-pain and WOMAC-function items loaded on similar factors. ICOAP-constant subscale had a large floor effect (33%).

Conclusions: P4 should be used to measure pain in patients with knee OA. It had acceptable measurement properties which is comparable to more widely used pain measures. WOMAC-pain shared a factorial structure with WOMAC-function indicating these measures might be capturing the same construct, questioning its validity to measure pain separately from function. ICOAP had acceptable properties. More work should compare pain measures in less severely affected OA populations.