Update of Markov Model on the Cost-effectiveness of Nonpharmacologic Interventions for Chronic Low Back Pain Compared to Usual Care
Published in: Spine, Volume 45, Issue 19, pages 1383–1385 (October 2020). doi: 10.1097/BRS.0000000000003539
Posted on RAND.org on December 16, 2020
Further validity test of a previously published model.
Summary of background data
The previous model was built using data from ten randomized trials and examined the 1-year effectiveness and cost-effectiveness of 17 nonpharmacologic interventions for chronic low back pain (CLBP), each compared to usual care alone. This update incorporated data from five additional trials.
Based on transition probabilities that were estimated using patient-level trial data, a hypothetical cohort of CLBP patients transitioned over time among four defined health states: high-impact chronic pain with substantial activity limitations; higher (moderate-impact) and lower (low-impact) pain without activity limitations; and no pain. As patients transitioned among health states, they accumulated quality-adjusted life-years, as well as healthcare and productivity costs. Costs and effects were calculated incremental to each study's version of usual care.
From the societal perspective and assuming a typical patient mix (25% low-impact, 35% moderate-impact, and 40% high-impact chronic pain), most interventions—including those newly added—were cost-effective (<$50,000/QALY) and demonstrated cost savings. From the payer perspective, fewer were cost-saving, but the same number were cost-effective. Results for the new studies generally mirrored others using the same interventions—for example, cognitive behavioral therapy (CBT) and physical therapy. A new acupuncture study had similar effectiveness to other acupuncture studies, but higher usual care costs, resulting in higher cost savings. Two new yoga studies' results were similar, but both differed from those of the original yoga study. Mindfulness-based stress reduction was similar to CBT for a typical patient mix but was twice as effective for those with high-impact chronic pain.
Markov modeling facilitates comparisons across interventions not directly compared in trials, using consistent outcome measures after balancing the baseline mix of patients. Outcomes also differed by pain impact level, emphasizing the need to measure CLBP subgroups.