low-value care, and we commend them on their research examining the clinical granularity of data that relies on diagnostic coding. Misclassifying a clinically appropriate service as lowvalue in our study 1 is an inherent limitation of using claims data. We would also note that misclassification can function in the other direction; that is, events excluded from the low-value category may be clinically inappropriate.We also agree that there are other important behavioral patient and clinician predictors of low-value care. Unfortunately, large-scale data sets that allow the statistical power to rigorously quantify predictors of low-value care do not contain the level of clinical information necessary to explore certain specific aspects of behavior. While our work cannot determine behavioral factors driving low-value care in detail, it suggests areas for targeting further analyses and interventions. 1 For example, research could examine at what point in clinical training low-value ordering habits begin, clinician characteristics associated with this, and the influences that reinforce this behavior.The work by Hahn and Gould pinpointing higher use of low-value breast cancer surveillance testing among clinicians who perceive that they had "nothing else to offer" highlights an especially pernicious habit. 2 When clinicians order testing that they know is unhelpful because they want to "do something," they miss the opportunity to communicate with and truly reassure a patient, and they ignore potential harms from downstream cascades of care. Beyond the cost implications of low-value care, these patient harms are the most concerning.Further research, including qualitative studies, should determine the modifiable clinician and patient behaviors that lead to low-value care. Unaddressed clinician behavioral factors can overwhelm the effectiveness of evidence-based clinical tools, even in countries where the financial and medico-legal incentives to order low-value care are less pronounced than in the United States. 3 At the same time, interventions to reduce low-value care that are designed with clinician behaviors in mind can be quite effective, 4 though few have been demonstrated at scale. We look forward to future work that will help inform large-scale interventions and steer the United States toward a more evidence-based and patient-centered care delivery system.