Development of a high-value care subscore on the internal medicine in-training examination.

Background: Although high-value care (HVC) that balances benefits of tests or treatments against potential harms and costs has been a recently emphasized competency for internal medicine (IM) residents, few tools to assess residents' knowledge of HVC are available.

Objective: To describe the development and initial results of an HVC subscore of the Internal Medicine In-Training Examination (IM-ITE).

Design: The HVC concepts were introduced to IM-ITE authors during question development. Three physicians independently reviewed each examination question for selection in the HVC subscore according to 6 HVC principles. The final subscore was determined by consensus. Data from the IM-ITE administered in October 2012 were analyzed at the program level.

Setting: U.S. IM residency programs.

Participants: 362 U.S. IM residency programs with IM-ITE data for at least 10 residents.

Measurements: Program-level performance on the HVC subscore was compared with performance on the overall IM-ITE, the Dartmouth Atlas hospital care intensity (HCI) index of the program's primary training hospital, and residents' attitudes about HVC assessed with a voluntary survey.

Results: The HVC subscore comprised 38 questions, including 21 (55%) on managing conservatively when appropriate and 14 (37%) on identifying low-value care. Of the 362 U.S. IM programs in the sample, 41% were in a different quartile when ranked based on the HVC subscore compared with overall IM-ITE performance. Rankings by HVC subscore and HCI index were modestly inversely associated, with 30% of programs ranked in the same quartile based on both measures.

Limitation: Knowledge of HVC assessed from examination vignettes may not reflect practice of HVC.

Conclusion: Although the HVC subscore has face validity and can contribute to evaluation of residents' HVC knowledge, additional tools are needed to accurately measure residents' proficiency in HVC.

Primary Funding Source: None.