Study Finds Huge Regional Variability in ‘Low-Value’ Orthopedic Services for Vets

So-called low-value orthopedic procedures, such as vertebroplasty for patients with asymptomatic spinal fractures, may be performed on veterans at dramatically different rates from one place to another, analysis of Medicare data showed.

In some hospital referral regions, certain low-value procedures were performed on one-quarter of Medicare-enrolled veterans at rates above 72 per 100 persons, while the same procedures were done elsewhere on hardly anyone, according to Samantha Auty, PhD, and colleagues at Boston University in Massachusetts.

However, high rates of one low-value service did not often correlate with high rates of other procedures in the same region, the authors reported in JAMA Network Open.

“These findings may inform procedure-specific strategies for reducing orthopedic LVS [low-value services], such as prior authorization or insurance network design based on procedure-specific utilization,” Auty and colleagues wrote. “Further research is needed to examine how procedure-specific patterns of orthopedic LVS translate into differences in overall spending and clinical outcomes to prioritize more targeted policy interventions.”

The problem of low-value medical diagnostics and treatments began to draw attention in 2010, kicking off a movement dubbed Less Is More and an American Board of Internal Medicine program called Choosing Wisely. Its point was that for many such services, the likely benefits are outweighed by the costs and risks.

In the orthopedic sphere, chronic pain is probably the most common focus for low-value procedures; the underlying causes can be hard to pin down even with advanced imaging, and surgical therapies may fail to bring relief and complications aren’t infrequent. Not all such services for chronic pain are low-value, of course — patient selection is key. For example, an MRI scan would usually be excessive for someone being seen for the first time with low back pain; the same is true for arthroscopy at an initial visit for uncomplicated knee pain.

Moreover, as with many disorders across medicine, specific approaches for musculoskeletal pain are more popular in some places than others, owing to local traditions and influential providers’ preferences. Thus, Auty and colleagues sought to examine a comprehensive dataset to see how low-value orthopedic services vary with geography, as a step toward better resource use.

That dataset was Medicare fee-for-service claims from 2017 to 2022 for veterans age 65 and up enrolled in the program. The researchers’ starting point was those beneficiaries who saw an orthopedic surgeon during this period. They measured claims for five different services considered low-value for the clinical conditions described in the records: vertebroplasty, spinal fusion, spinal injections, advanced spinal imaging, and knee arthroscopy.

Importantly, an order or claim for one of these was categorized as low-value only when the patients’ clinical data, as contained in the records, corresponded with published definitions of low-value care. As an example, vertebroplasty is appropriate when a vertebral compression fracture is unhealed and causing pain, and conservative treatments such as rest and analgesic medications have been tried without success. Absent these factors, vertebroplasty is likely to be unnecessarily aggressive.

Claims for such procedures were compiled for the 306 hospital referral regions defined in Medicare. Rates were calculated per 100 enrollees in each region as 25th-75th percentiles. Auty and colleagues reported these data in map form, with four colors (pale pink to dark red) to represent very low, low, high, and very high rates. These were as follows for the five types of procedures:

• Vertebroplasty: very low, 0.0-3.4; low, 3.4-6.8; high, 6.8-11.6; very high, 11.6-42.1
• Spinal fusion: very low, 0.0-1.0; low, 1.0-1.5; high, 1.5-2.4; very high, 2.4-9.6
• Spinal injection: very low, 0.0-1.7; low, 1.7-3.3; high, 3.3-5.8; very high, 5.8-35.7
• Back imaging: very low, 9.1-36.8; low, 36.8-43.6; high, 43.6-50.7; very high, 50.7-71.9
• Knee arthroscopy: low, 0.0-0.1; low, 0.1-0.3; high, 0.3-0.4; very high, 0.4-2.3

Perhaps the most noteworthy aspect of these figures was that the 75th percentile in the “very high” categories was massively higher than the 75th percentile in the “very low” categories, indicating dramatic variation in low-value service rates.

The maps themselves were somewhat misleading, however: the hospital referral regions vary tremendously in land area, with those in the western U.S. covering huge areas while East Coast regions containing many more enrollees were almost invisibly small. Thus, colorations in those western regions tended to dominate the visual representation. (Use of Mercator projections for the maps, which exaggerate the size of northern areas, further distorted the impression.)

But another finding rendered this defect less meaningful: Auty and colleagues found little correlation between high rates of low-value services in a given region. That is, a high rate for one procedure type did not predict high rates of others. The researchers calculated correlation coefficients among service types, with r values ranging from -0.25 to 0.22, with most clustering between -0.5 and 0.15. Overall this finding suggested “that regional variation in individual orthopedic LVS may be largely procedure-specific rather than driven by regional utilization tendencies,” the researchers wrote.

Ultimately, the study highlighted the need for more research to identify factors driving low-value orthopedic care in the regions with high rates, which could then suggest policies and interventions to bring them down.

An important limitation to the study was that the focus on veterans meant that enrollees were predominantly men, and thus the findings may not be generalizable to patients seen in the broader community.