Chronic pain is common, yet treatments often fall short

Prescribing an effective therapy for chronic pain can prove challenging for doctors. That’s because nearly everything in patients’ lives — from their health history to sleep patterns to job stress — contributes to how they experience pain.  

Recently published study findings bear out this difficulty by showing that many common pain treatments perform only slightly better than a placebo. The authors suggested that a whole-person approach may deliver more effective pain care. In the long run, tools such as artificial intelligence might someday help doctors identify risk factors and design more individualized pain treatments. 

Researchers from Northwestern University, Johns Hopkins University and the University of Washington School of Medicine reviewed human clinical studies on factors linked to outcomes after surgical and nonsurgical procedures for chronic pain.  

Their paper, which appeared July 7 in BMJ Medicine, included systematic reviews and clinical studies published through March 2025. 

Chronic pain affects an estimated 20% to 34% of people worldwide and is a leading cause of disability. 

“Pain is often thought of as a uniform symptom,” said co-author Dr. Yian Chen, associate professor of anesthesiology and pain medicine at the University of Washington’s School of Medicine. “But pain is caused by a complex web of factors, and pain itself is often classified in different ways.” 

 For some patients, finding the source of pain is relatively simple. Localized pain that has a single cause, such as arthritis, can often be identified through routine tests. Doctors can then assess a patient's overall health and risk factors to come up with a tailored approach to care. But what about patients experiencing more mysterious pain?  

 “With ailments such as cancer, specific biomarkers exist that allow doctors to design treatments based on what they know will be most effective for that specific person,” Chen said. When it comes to pain, however, no such markers exist.  

 The research team analyzed a range of risk factors associated with better or worse outcomes after procedures aimed at relieving pain. Risk factors often overlap and amplify one another, a phenomenon highlighted by the review. For example, depression and pain contribute to sleep problems, and sleep problems also can contribute to pain and depression. Of particular importance to patient outcomes are: 

• Sleep problems: Preexisting sleep disorders are strongly linked with poorer treatment results and higher risk of persistent post-surgical pain. 
• Obesity: Associated with multiple pain syndromes and, in some contexts, higher complication and failure rates. 
• Smoking: Independently tied to small negative effects on procedural outcomes, although overlapping risk factors are common. 
• Opioid use: Linked to worse surgical and nonsurgical outcomes and higher postoperative opioid needs. 
• Sensitization: Heightened nervous system activity is strongly associated with multiple pain conditions, more severe symptoms, and failure of treatments. 

By leveraging tools such as large language models, Chen and his colleagues hope researchers can develop new tools that uncover what’s causing pain in patients and the most effective treatments.  

“I think that there’s a balance that exists between payers and providers, with patients in the middle,” Chen said. “The best way for providers to ensure that their patients are receiving the best care is to pinpoint the interventions that work within specific populations. Otherwise, we risk applying ineffective and expensive procedures across the board.” 

More research is needed, however, to ensure that large language models draw the right conclusions from the existing research and patient-specific information.   

“There is a lot of heterogeneity in study design and patient population, which raises important questions about the best way to compile, integrate and understand the existing data,” Chen said. “We need to be able to weigh the relative impact of each risk factor before we can rely on artificial intelligence to draw sound conclusions.” 

Drs. Eric J. Wang from Johns Hopkins Medical Institutions, Alex Roybal from Northwestern University, and Steven P. Cohen from Northwestern and Walter Reed National Military Medical Center, also contributed to this study.  

The research was funded in part by a grant from MIRROR, Uniformed Services University of the Health Sciences, U.S. Department of Defense (HU00011920011). 

For details about UW Medicine, please visit http://uwmedicine.org/about.