Background and Objectives: Pain is one of the most important outcome variables in epidemiologic studies of musculoskeletal disorders and is most commonly assessed retrospectively. This study examined the reliability/validity of retrospective pain ratings.

Methods: 400 individuals reporting low back pain (LBP) in a population-based survey were invited to participate in a year-long study of the course of LBP. At the start and end, they completed a questionnaire (demographics, symptoms, function, Chronic Pain Grade Questionnaire (CPGQ)). Each week, they completed a one-page diary (numeric rating scale for pain intensity, work limitations and social limitations due to LBP). Participants could choose between electronic or paper versions. Intraclass correlation coefficients indicated the agreement between the weekly ratings and the corresponding retrospective assessment (CPGQ).

Results: 348 individuals agreed to participate; 250 provided complete data-sets. Agreement between prospective and retrospective assessments was good (ICC pain=0.73, 95%CI 0.65–0.79; ICC work limitations=0.82, 95%CI 0.77–0.86; ICC social limitations=0.78, 95%CI 0.71–0.83). Separate analyses revealed no influence on the ICCs of completion-method, age and sex, but an influence of pain severity (severe: ICCs = 0.73, 0.88, 0.82, respectively; mild/moderate: 0.19, 0.22, 0.37, respectively) and pain duration (acute: 0.75, 0.81, 0.79, respectively; sub-acute/chronic: 0.52, 0.77, 0.57, respectively).

Conclusion: The results have important implications for clinical practice and epidemiological studies. For clinical practice, where patients with pain are being treated, retrospective pain assessments up to 6 months seem to be reliable. However, this is not the case in epidemiological studies in which many individuals with no or only moderate pain are involved.