Highlight
- Physicians’ performance on the Longitudinal Knowledge Assessment (LKA) inversely correlates with the provision of low-value services (LVSs) in older Medicare patients.
- Top quartile LKA performers have a 7.9% lower adjusted likelihood of ordering any LVS compared to bottom quartiles, with notable reductions in diagnostic testing, cancer screening, and imaging.
- Specific LVSs such as unnecessary triiodothyronine testing and prostate-specific antigen screening in elderly men show the most pronounced differences related to physician knowledge.
Study Background
Low-value services—tests, procedures, or treatments that provide little or no benefit to patients—remain a substantial driver of inefficiency in U.S. healthcare, contributing to wasteful spending without improving outcomes. This is particularly relevant in older adults, a population with high healthcare utilization and multiple comorbidities, for whom unnecessary interventions can increase harm risk. Understanding factors that influence LVS use is critical to improving care quality and system sustainability.
Study Design
This cross-sectional study analyzed data from general internists participating in the American Board of Internal Medicine’s (ABIM) Longitudinal Knowledge Assessment (LKA) during 2022 or 2023. The cohort included physicians caring for Medicare fee-for-service beneficiaries aged 67 years or older in outpatient settings during their first LKA assessment year.
Physician exposure was defined using their first-year LKA scores, which reflect up-to-date clinical knowledge regarding internal medicine. The primary outcome measured was the receipt of any one of 25 low-value primary care services, including diagnostic tests, preventive screenings, and imaging studies. Patient-level regression analyses adjusted for demographic variables, comorbid conditions, physician practice characteristics, and patient location to isolate the association between physician knowledge and LVS use.
Key Findings
The study included 898,365 patients and 7,089 physicians. Among patients, 59% were female with a mean age of 76.8 years; 30% received at least one LVS during the study year. Physicians averaged 52.5 years of age, with a nearly even gender distribution.
Patients managed by physicians with LKA scores in the top quartile exhibited a 7.9% lower adjusted likelihood of receiving any low-value service compared to those with physicians in the bottom quartile (31.0% vs. 28.6%). Although the absolute difference in LVS exposure was modest (-2.5 percentage points), it was statistically significant (P < .001), indicating that higher physician knowledge corresponds with reductions in potentially unnecessary care.
When stratified by LVS subtype:
– Diagnostic or preventive testing: 10.4% (bottom quartile) vs. 8.7% (top quartile); absolute difference -1.7 percentage points (P < .001).
– Cancer screening (e.g., prostate-specific antigen tests in elderly males): 13.8% vs. 12.3%; absolute difference -1.5 percentage points (P < .001).
– Imaging studies: 13.8% vs. 13.2%; absolute difference -0.6 percentage points (P = .02).
Notably, the greatest knowledge-related differences were observed in total and free triiodothyronine testing for hypothyroid patients and prostate-specific antigen screening in men aged 75 years and older—services commonly deemed low-value due to low clinical yield and potential for harm in these populations.
Expert Commentary
The association between higher clinical knowledge and reduced LVS use underscores the importance of ongoing physician education in curbing unnecessary healthcare. As measured by the ABIM LKA, physicians maintaining current knowledge seem better positioned to apply evidence-based guidelines and avoid interventions lacking clinical benefit.
However, the modest absolute differences suggest that while physician knowledge is an important factor, other influences—such as patient preferences, systemic incentives, and local practice patterns—also play roles in LVS utilization. Additionally, the cross-sectional design precludes definitive causal inference, and residual confounding by unmeasured variables cannot be excluded.
Further studies could explore longitudinal changes in LVS use relative to evolving physician knowledge and evaluate interventions aimed at improving knowledge as a strategy to reduce low-value care. Incorporating patient-centered decision aids alongside physician education could enhance efforts to optimize testing and treatment appropriateness.
Conclusion
This study demonstrates that general internists with higher scores on the Longitudinal Knowledge Assessment are less likely to provide low-value services to older Medicare beneficiaries. These findings highlight physician clinical knowledge as a potentially modifiable factor to improve care quality and reduce wasteful health expenditures. Reinforcing continuous professional development and evidence-based practice adherence could be key strategies in tackling overuse in clinical care.
Funding and Registration
Details regarding study funding and clinical trial registration were not reported as this was an observational study using existing administrative and assessment data.
References
1. Vandergrift JL, Landon BE, Weng W, Gray BM. Low-Value Services and Longitudinal Knowledge Assessment Performance. JAMA Intern Med. 2026 Jul 13. PMID: 42440313.
2. ABIM. Longitudinal Knowledge Assessment Program Details. American Board of Internal Medicine. https://www.abim.org.
3. Schwartz AL, Landon BE, el al. Measuring Low-Value Care. JAMA Intern Med. 2020;180(2):284-285.
4. Choosing Wisely Campaign. Reducing Low-Value Care. https://www.choosingwisely.org.
