Major Finding: A single end-of-day pain score correlated with multiple pain scores measured during the day by a coefficient of 0.88 on weekdays and 0.86 on weekend days.
Data Source: Visual analog scale scores recorded on weekdays and weekend days by 157 patients with osteoarthritis.
Disclosures: Study supported by American College of Rheumatology Research and Education Foundation. The investigators reported no other conflicts of interest.
Patients with osteoarthritis who are asked just before bedtime to estimate their average pain during the day are able to produce accurate estimates, meaning that clinicians can rely on these end-of-day estimates without having to worry that the patient's recall may be biased.
These findings differ from those in other studies that have suggested that end-of-day recall might be affected by more recent pain readings, by whether it was a weekday or a weekend, or by other patient characteristics.
The patients in the current study all had osteoarthritis (OA) of the hand, hip, or knee. Their mean age was 62 years, and women accounted for 52% of the study cohort, wrote by Kelli D. Allen, Ph.D., and colleagues at Duke University, Durham, N.C. (J. Pain 2009 Jan. 22 [doi:10.1016/j.jpain.2009.09.007]).
They were each given a handheld computer that reminded them approximately every 2 hours to record their current level of pain on a visual-analog scale. At the end of the day, the computer also prompted the patients to estimate their average pain over the entire day. Each patient recorded pain estimates on 2 days, once on a weekday and once on a weekend day.
On average, pain scores recorded throughout the day on the 1-100 visual-analog scale were 35 on weekdays and 33 on weekends.
Patients' average end-of-day recalled pain scores were 37 on weekdays and 34 on weekends. None of the differences were statistically significant.
The overall correlation coefficient between end-of-day estimates and the average of the seven estimates during the day was 0.88 on weekdays and 0.86 on weekend days, indicating a high–and statistically significant–degree of correlation.
The investigators were able to eliminate the possibility that the patients may have been biased by their most recent pain reading when recalling their average pain. They did this by calculating the patient's average pain after deleting the final reading of the day and repeating the correlation analysis. If that reading affected the patient's recall significantly, the investigators would have seen a substantial change in the correlation coefficients. In fact, the correlation coefficients were almost identical–0.87 on weekdays and 0.85 on weekends.
In addition, the investigators found that none of the patient characteristics they tested had a significant correlation with the accuracy of the end-of-day estimates. In particular, they found that age, gender, race, joint site, and the tendency of the patient to catastrophize pain (measured by the Coping Strategies Questionnaire) had no significant interactions with accuracy.
The investigators acknowledged that pain recall over a 1-week period may be less accurate than 1-day recall. And they pointed out that they demonstrated the accuracy of patient recall only for the chronic pain of OA, which may account for the far lower degree of recall accuracy than other investigators saw in a study of acute pain in postsurgical patients.