Exploring the Prevalence and Construct Validity of High-Impact Chronic Pain Across Chronic Low-Back Pain Study Samples

Published in: The Spine Journal (2019). doi: 10.1016/j.spinee.2019.03.00

Posted on RAND.org on April 11, 2019

by Patricia M. Herman, Nicholas Broten, Tara Lavelle, Melony E. Sorbero, Ian D. Coulter

Read More

Access further information on this document at Elsevier Inc

This article was published outside of RAND. The full text of the article can be found at the link above.

Background Context

The US National Pain Strategy focused attention on high-impact chronic pain and its restrictions. Although many interventions have been studied for chronic low-back pain, results are typically reported for heterogeneous samples. To better understand chronic pain and target interventions to those who most need care, more granular classifications recognizing chronic pain's impact are needed.


To test whether chronic pain impact levels can be identified in chronic low-back pain clinical trial samples, examine the baseline patient mix across studies, and evaluate the construct validity of high-impact chronic pain.

Study Design/Setting

Descriptive analyses using twelve large study datasets. Patient samples. Chronic low-back pain patients in non-surgical, non-pharmacologic trials in the US, Canada and UK.Outcome measures. Preference-based health utilities from the SF-6D and EQ-5D, employment status and absenteesim.


We used two logistic regression models to predict whether someone had high-impact chronic pain, and whether the remainder had low- or moderate-impact chronic pain. We developed these models using two datasets and models with the best predictive power were used to impute impact levels for six other datasets. Stratified by these estimated chronic pain impact levels, we characterized the case mix of patients at baseline in each dataset, and summarized their health-utilities, and work productivity. This study was funded by a National Center for Complementary and Integrative Medicine grant. The authors have no potential conflicts of interest.


The logistic models had excellent predictive power to identify those with high-impact chronic pain. Although studies were all of chronic low-back pain patients, the baseline mix of patients varied widely. Across all datasets, utilities and productivity were similar for those with high-impact chronic pain, and worsened as chronic pain impact increased.


There is a need to better categorize chronic pain patients to allow the targeting of optimal interventions for those with each level of chronic pain impact.

Research conducted by

This report is part of the RAND Corporation external publication series. Many RAND studies are published in peer-reviewed scholarly journals, as chapters in commercial books, or as documents published by other organizations.

The RAND Corporation is a nonprofit institution that helps improve policy and decisionmaking through research and analysis. RAND's publications do not necessarily reflect the opinions of its research clients and sponsors.