SMART Study Compares Staged Nonpharmacologic Care for Chronic Low Back Pain

Key Takeaways

• Physical therapy was associated with an early functional advantage over cognitive behavioral therapy at 10 weeks.
• Pain intensity did not differ at 10 weeks, and no stage II strategy separated from another at 52 weeks among nonresponders.
• The functional gap was smaller than the minimum important difference, and interpretation was tempered by lower treatment initiation, lack of blinding, and pandemic-related sample-size reduction.

Physical therapy was linked to greater 10-week functional improvement than cognitive behavioral therapy in the SMART trial in Annals of Internal Medicine, with a 2.8-point adjusted mean difference on the Oswestry Disability Index. The staged comparison enrolled 749 adults with chronic low back pain across three health care systems in a sequential multiple-assignment randomized trial, with nonresponders re-randomized after stage I and followed for 52 weeks. Longer-term findings did not differ across the stage II strategies among nonresponders.

SMART was a multisite sequential multiple-assignment randomized trial across three health care systems in 749 adults with chronic low back pain. Stage I consisted of 8 weeks of physical therapy or cognitive behavioral therapy, and nonresponders were randomly assigned again to 8 weeks of stage II treatment. The second-stage comparison evaluated switching treatments versus mindfulness. Co-primary outcomes were function on the 0-to-100 Oswestry Disability Index and pain intensity on a 0-to-10 scale at 10, 26, and 52 weeks.

At 10 weeks, physical therapy was associated with greater functional improvement than cognitive behavioral therapy, with an adjusted mean ODI difference of 2.8 and a 96% CI of 0.38 to 5.1. That difference was below the minimum important difference of 6. Pain intensity did not differ between the stage I groups at 10 weeks, with an adjusted mean difference of 0.32 and a 99% CI of negative 0.07 to 0.71. The early difference was seen in function rather than pain.

Among nonresponders, the second-stage strategies did not differ for either function or pain intensity at 52 weeks. The adjusted mean difference for function was 0.43 on the ODI, with a 96% CI from negative 0.29 to 2.4. The adjusted mean difference for pain intensity was negative 0.05, with a 96% CI from negative 0.58 to 0.48. The longer-term adaptive comparisons did not show separation on the co-primary outcomes.

Authors noted several limitations. Treatment initiation was lower than expected, particularly for cognitive behavioral therapy and for nonresponders. Participants were not blinded, and the sample size was reduced because of the COVID-19 pandemic. Overall, the trial showed a modest early functional advantage with physical therapy, while response-guided second-stage options did not differ at 52 weeks among nonresponders.