New analysis shows use and predictors of low-value care in health systems nationwide

Researchers estimate that up to $101 billion in health care spending is wasted each year due to over-treatment or the delivery of “low-value care.” Low-value care includes a wide array of tests and treatments that are medically unnecessary and for which the potential for harm outweigh the potential for benefit. While researchers have charted the ongoing use of low-value care on national and regional levels, little research exists on how health systems across the country use low-value care and how they compare to each other. A new study by investigators from Brigham and Women’s Hospital and the Dartmouth Institute for Health Policy and Clinical Practice examined the use of 41 low-value services at 556 U.S. health systems. Their results, published in JAMA Internal Medicine, map hotspots of low-value care use at individual health systems and shed light on predictors of this use.

“More and more Americans are getting care from health systems as opposed to standalone practices. Given actionable data, these systems have enormous potential to influence decision-making on low-value care,” said lead author Ishani Ganguli, MD, MPH, a researcher in the Brigham’s Division of General Internal Medicine and Primary Care and Assistant Professor of Medicine at Harvard Medical School. “We hope this work might motivate systems to measure and intervene on low-value care internally.”

The researchers found that preoperative laboratory testing in healthy patients before low-risk surgeries, prostate cancer screening in men over 70 years of age and use of antipsychotic medications in patients with dementia were the most common forms of low-value care among those studied. The preoperative lab tests, for example, are not recommended because they do not improve surgical outcomes and can show false alarms, among other issues.

The researchers identified an array of factors associated with health systems whose patients received more low-value care. The systems tended to have a smaller share of primary care physicians, no associated teaching hospital, headquarters located in the South or West (compared to the Northeast or Midwest), and proportionally more patients who were racial and ethnic minorities. The use of low-value care was also correlated with more overall health care spending in the area.

To conduct their study, the authors examined national claims data for Medicare beneficiaries over the age of 65 and linked each one to a health system based on where they received the plurality of their primary care. Drawing upon prior definitions of 41 low-value services, the researchers measured the use of each of these services among patients eligible for the given service. Then, they combined the 28 most common low-value services to create composite low-value care scores to compare systems.

The researchers note limitations of their study. For instance, Medicare claims data do not have the clinical details to confirm why a physician might have ordered a certain test or procedure for a patient. And the estimates capture just a snapshot of low-value services in a specific time period. Nonetheless, they hope this work might help health systems intervene on low-value care use, for example through employee education, clinical decision support systems, changing workplace culture, or adjusting reimbursement models.

Conflict of Interest: Dr. Ganguli reports receiving grants from the Agency for Healthcare Research and Quality, the National Institute of Aging, the Robert Wood Johnson Foundation, and International Business Machines Inc; and personal fees from Haven, FPrime, and Blue Cross Blue Shield Massachusetts that are unrelated to this work.

Funding: Funding for this work was provided by Arnold Ventures (19-02548) and the Agency for Healthcare Research and Quality’s Comparative Health System Performance Initiative (1U19HS024075).

Paper cited: Ganguli I et al. “Low-Value Care at the Actionable Level of Individual Health Systems” JAMA Internal Medicine DOI: 10.1001/jamainternmed.2021.5531