Can we assess baseline pain and global health retrospectively?

To study the agreement between patients' actual baseline assessments of pain and global health before treatment and retrospective assessments collected 2 weeks after treatment.

Data were collected in a prospective study of 200 rheumatology outpatients treated with a local corticosteroid injection. At baseline and 2-week follow-up, localized pain and global health were assessed on 100 mm visual analogue scales. The follow-up questionnaire was extended with a retrospective assessment of pain and global health before treatment.

At follow-up patients slightly overestimated the severity of pain and global health before treatment. Actual and retrospective assessments were adequately correlated (pain: r(r(r(s) = 0.73; global health: r(s) = 0.67). Bland-Altman analysis showed that both pain and global health were characterized by high intra-individual variation between actual and retrospective assessments, with the 95% limits of agreement (-37.3 to 32.3 mm for pain and -49.7 to 37.8 mm for global health) far exceeding proposed values for minimal clinically important differences.

Over a 2-week interval, patients' retrospective assessments of baseline pain and global health are fairly accurate and adequately correlated with actual baseline scores. At the group level, retrospective assessments can provide acceptable data on baseline pain and global health. The wide variability between actual and retrospective assessments, however, indicates that even over short time intervals there is poor individual agreement between the two methods.