Pain assessment on a numerical scale with uncertainty intervals: a proof-of-concept simulation study.

Abstract

Background: Reliable and validated scores assessing pain-related outcomes are an essential component of pain management. Point estimates, e.g., on the numeric rating scale (NRS), are widely used. Given the broad spectrum of physiological and psychological factors involved in a patient's pain experience, these point estimates entail inherent uncertainty. To account for this uncertainty, we propose a statistical framework featuring uncertainty intervals on a numerical scale assessing pain intensity.

Methods: We describe a non-parametric statistical method to estimate the effectiveness of a pain intervention when patients provide an uncertainty interval of pain intensity rather than a single point estimate. We consider pain intensities on a generic numerical pain scale (NPS) ranging from 0 to 10 and illustrate the method's performance with proof-of-concept simulation studies and sensitivity analyses.

Results: The simulation studies demonstrate that the non-parametric method can derive correct estimates of the average treatment effects in idealized settings. Importantly, the method can represent the traditional pain assessment with point estimates when the widths of the uncertainty intervals are gradually decreased toward the mean of the uncertainty interval.

Conclusion: We proposed a new statistical framework to account for patient-specific uncertainties in pain intensity as measured on a numerical scale. The clinical importance of the method lies in its ability to reflect the large heterogeneity of individual pain experiences and the possibility of investigating pain-related aspects that go beyond a traditional pain assessment with point estimates. Future clinical studies are required to assess the method's clinical validity and utility.