Related Experiment Video
Updated: May 9, 2025
09:38
Determining Pain Detection and Tolerance Thresholds Using an Integrated, Multi-Modal Pain Task Battery
Published on: April 14, 2016
12.5K
Estimating controlled direct treatment effects on pain intensity using structural mean models: application to pain
Rui Wang1, Patrick J Heagerty1,2, Kwun Chuen Gary Chan1
1Department of Biostatistics, University of Washington, Seattle, USA.
Medrxiv : the Preprint Server for Health Sciences
|April 29, 2025
Summary
New causal inference methods accurately account for concurrent analgesic use in pain randomized controlled trials (RCTs). These advanced techniques may reveal larger treatment effects for interventions like epidural steroid injections (ESI).
Area of Science:
- Clinical Epidemiology
- Biostatistics
- Pain Management Research
Background:
- Traditional analyses in pain randomized controlled trials (RCTs) often underutilize methods accounting for concurrent analgesic use.
- Failure to incorporate concurrent analgesic use can diminish estimated treatment effects in primary pain analyses.
- Epidural steroid injection (ESI) is a common treatment for leg pain intensity (LPI).
Purpose of the Study:
- To reanalyze RCT data for epidural steroid injection (ESI) using contemporary causal inference methods that account for concurrent analgesic use.
- To define and estimate an 'attributable to ESI estimand' representing the controlled direct effect of ESI.
- To compare treatment effect estimates using traditional methods versus causal inference approaches.
Main Methods:
- Employed contemporary causal inference methods, specifically structural mean models (SMMs), to reanalyze existing RCT data.
- Defined a composite pain intensity outcome, the QPAC1.5, and utilized SMMs (estimating equations [EE], g-estimation, generalized method of moments [GMM]) to estimate the target estimand.
- Compared results from SMMs with traditional strict intention-to-treat (strict ITT) analysis on leg pain intensity (LPI) measured by the numeric rating scale (NRS).
Main Results:
- The strict ITT analysis yielded an ESI treatment effect of -0.751 NRS points (95% CI: [-1.287, -0.214]).
- SMM-based estimates for the attributable to ESI estimand were: EE -0.864 (95% CI: [-3.207, 1.478]), g-estimation -0.935 (95% CI: [-1.779, 0.090]), GMM -0.653 (95% CI: [-1.218, -0.089]), and QPAC1.5 -0.930 (95% CI: [-1.508, -0.352]).
- Causal inference methods accounted for analgesic use without assuming sequential ignorability, potentially yielding larger treatment effects.
Conclusions:
- Contemporary causal inference methods offer a valid approach to account for concurrent analgesic use in pain RCTs.
- These methods, applied to ESI for LPI, provide alternative estimands that may yield different, potentially larger, treatment effect estimates compared to strict ITT.
- The study highlights the utility of advanced statistical techniques for more accurate treatment effect estimation in pain research.
