Self-reports of pain intensity and direct observations of pain behavior: when are they correlated?

Meta-analytic techniques were utilized to investigate the relationship between self-reports of pain intensity and direct observations of pain behavior. Estimation of the overall effect size from 29 studies and 85 effect sizes yielded a moderately positive association, z=0.26. High variability across studies permitted a random-effects moderator analysis that determined chronicity of pain, the timing of the pain assessment, the use of global measures of pain behavior, and pain site significantly moderate the relationship between self-reports of pain intensity and direct observations of pain behavior. These findings indicate that self-reports of pain intensity and direct observations of pain behavior are more likely to be significantly related to each other when the individual being studied has acute pain (z=0.35), when the self-report of pain intensity data are collected soon after the observation of pain behavior (z=0.40), when global composite measures are used to quantify pain behavior (z=0.37), and when the person being observed suffers from chronic low back pain (z=0.30). Other factors not found to be significant moderators include: extent of observer training, relevance of the pain-inducing task, and pain behavior observation measure used. The implications of the findings for the assessment of pain are discussed.