Harmonising knee pain patient-reported outcomes: a systematic literature review and meta-analysis of Patient Acceptable Symptom State (PASS) and individual participant data (IPD).
Georgopoulos V., Smith S., McWilliams DF., Steultjens MPM., Williams A., Price A., Valdes AM., Vincent TL., Watt FE., Walsh DA.
OBJECTIVE: In order to facilitate data pooling between studies, we explored harmonisation of patient-reported outcome measures (PROMs) in people with knee pain due to osteoarthritis or knee trauma, using the Patient Acceptable Symptom State scores (PASS) as a criterion. METHODS: We undertook a systematic literature review (SLR) of PASS scores, and performed individual participant data (IPD) analysis of score distributions from concurrently completed PROM pairs. Numerical rating scales (NRS), visual analogue scales, KOOS and WOMAC pain questionnaires were standardised to 0 to 100 (worst) scales. Meta-regression explored associations of PASS. Bland Altman plots compared PROM scores within individuals using IPD from WebEx, KICK, MenTOR and NEKO studies. RESULTS: SLR identified 18 studies reporting PASS in people with knee pain. Pooled standardised PASS was 27 (95% CI: 21 to 35; n = 6,339). PASS was statistically similar for each standardised PROM. Lower PASS was associated with lower baseline pain (β = 0.49, P = 0.01) and longer time from treatment initiation (Q = 6.35, P = 0.04). PASS scores were lowest in ligament rupture (12, 95% CI: 11 to 13), but similar between knee osteoarthritis (31, 95% CI: 26 to 36) and meniscal tear (27, 95% CI: 20 to 35). In IPD, standardised PROMs each revealed similar group mean scores, but scores within individuals diverged between PROMs (LoA between -7 to -38 and +25 to 52). CONCLUSION: Different standardised PROMs give similar PASS thresholds in group data. PASS thresholds may be affected more by patient and treatment characteristics than between PROMs. However, different PROMs give divergent scores within individuals, possibly reflecting different experiences of pain.