Is “Percent Pain Reduction” a Valid Metric of Clinical Pain Improvement?
Advisor(s)
Dr. A. V. Apkarian; Northwestern University, Feinberg School of Medicine
Andrew Vigotsky; Northwestern University, Departments of Biomedical Engineering and Statistics
Discipline
Medical and Health Sciences
Start Date
21-4-2021 10:45 AM
End Date
21-4-2021 11:05 AM
Abstract
Chronic pain is the most prevalent health condition in the United States, affecting over 116 million Americans, and therefore is the focus of many clinical trials. Studies on chronic pain treatment commonly report improvements in pain as a percent reduction from an initial value. Importantly percent reduction implies that improvements in pain are multiplicative. Although percent reductions are conceptually simple, empirically, it is unclear whether changes are truly multiplicative in nature. We assessed the validity of this assumption using longitudinal data from multiple randomized controlled trials. In each dataset, we assessed the presence of two hallmarks of a multiplicative effect: (1) whether the decrease in pain scales with initial pain; and (2) whether the residual error scales with greater pain ratings. The data did not meet either of these conditions. Since (1) changes in pain did not correlate with pre-intervention pain ratings and (2) residual error did not scale with post-intervention pain, pain reductions do not exhibit multiplicative properties. Instead, the data appear additive rather than multiplicative. Thus, reporting percent reductions in pain may be misleading. Instead, researchers and clinicians should report differences in pain, which more appropriately, represents the nature of changes in clinical pain.
Is “Percent Pain Reduction” a Valid Metric of Clinical Pain Improvement?
Chronic pain is the most prevalent health condition in the United States, affecting over 116 million Americans, and therefore is the focus of many clinical trials. Studies on chronic pain treatment commonly report improvements in pain as a percent reduction from an initial value. Importantly percent reduction implies that improvements in pain are multiplicative. Although percent reductions are conceptually simple, empirically, it is unclear whether changes are truly multiplicative in nature. We assessed the validity of this assumption using longitudinal data from multiple randomized controlled trials. In each dataset, we assessed the presence of two hallmarks of a multiplicative effect: (1) whether the decrease in pain scales with initial pain; and (2) whether the residual error scales with greater pain ratings. The data did not meet either of these conditions. Since (1) changes in pain did not correlate with pre-intervention pain ratings and (2) residual error did not scale with post-intervention pain, pain reductions do not exhibit multiplicative properties. Instead, the data appear additive rather than multiplicative. Thus, reporting percent reductions in pain may be misleading. Instead, researchers and clinicians should report differences in pain, which more appropriately, represents the nature of changes in clinical pain.