In clinical hand osteoarthritis research, self-reported pain questionnaires do not reflect the patient experience

Pain in hand osteoarthritis (OA) is evaluated with repeated pain questionnaires. It is unclear whether these questionnaires adequately capture changes in pain recalled by patients. This study investigated whether changes on pain questionnaires (real-time evaluation) correspond to recalled pain. Data from hand OA patients from the HOSTAS cohort (four one-yearly) and HOPE trial (one six-week interval) were used. Pain was measured with the Australian/Canadian hand Osteoarthritis Index (AUSCAN, range 0-20) and a recall question (how is the pain compared with your last visit). Changes in AUSCAN pain were categorized into improved (≤-1), stable or worsened pain (≥1) and compared with the recall question using Cohen's kappa and percentage agreement. We determined concordance between measurement methods, and investigated associations of mental wellbeing and illness perceptions with concordance using GEE. Of 708 intervals from HOSTAS (307 patients, 82% women, mean age 61.0 years, mean AUSCAN 9.1), AUSCAN changes and recall were concordant in 42% (Cohen's kappa 0.13). There was concordance in 47% of 86 intervals (Cohen's kappa 0.14) from the HOPE trial (86 patients, 80% women, mean age 63.5, mean AUSCAN 10.7). The most frequent recall answer was worsened pain in the HOSTAS (60%), improved pain in the HOPE trial (76%). In both studies, AUSCAN pain most frequently improved. Depression and anxiety showed no association with concordance. Changes in repeatedly measured AUSCAN pain often differ from the recalled course of pain over the same period. This has profound implications for evaluating patient reported pain in clinical trials.