Do Nurse Practitioners Deliver Care as Efficiently and Effectively as Physicians in Emergency Settings?

Chan and Chen (2025) examine how the productivity of nurse practitioners compares to physicians in emergency departments. They study 1.1 million patient visits from Veterans Health Administration emergency departments, using quasi-random assignment of patients to providers. They find that nurse practitioners use more resources, increasing length of stay by about 11 percent and costs by 7 percent. They also raise 30-day preventable hospitalizations by roughly 20 percent, with no significant effect on mortality. However, they show that performance differences shrink for less complex cases and that there is substantial overlap in productivity across providers within both professions.

Why This Article Was Selected for The Policy Scientist

This article addresses a central policy question: how to allocate skilled labor in health care systems facing persistent provider shortages and rising costs. The role of nurse practitioners has expanded rapidly, making credible evidence on comparative productivity essential for workforce design, scope-of-practice laws, and cost containment strategies. Chan and Chen, who have contributed extensively to the literature on provider behavior and health care productivity, offer timely evidence as policymakers reconsider staffing models. The study leverages a large administrative dataset from the Veterans Health Administration and applies a quasi-experimental instrumental variables design, strengthening causal interpretation relative to standard regression approaches. While the setting may limit generalizability to non-integrated systems, the methods and scale represent a meaningful empirical advance.

Full Citation and Link to Article

Chan, D., & Chen, Y. (forthcoming). The productivity of professions: Evidence from the emergency department. American Economic Review. https://doi.org/10.1257/aer.20241007

Central Research QuestionThis study asks how the productivity of nurse practitioners compares to that of physicians when both perform overlapping clinical tasks in emergency departments. Productivity is defined in a multidimensional sense, encompassing resource utilization—such as length of stay and treatment costs—as well as downstream patient outcomes, including preventable hospitalizations and mortality. The central objective is to isolate the causal effect of provider type on these outcomes, while accounting for patient selection and case complexity. A related question concerns the extent to which differences across professions reflect systematic training and selection versus variation within professions. The paper also examines whether productivity differences depend on patient severity and whether substantial overlap exists in performance distributions across the two provider groups.

Previous LiteratureThe paper situates itself within several strands of literature, including research on occupational licensing, health care workforce composition, and provider productivity. Earlier studies comparing nurse practitioners and physicians have largely focused on primary care settings and often relied on small samples or research designs with limited causal identification. These studies frequently found minimal differences in outcomes, though their statistical power and external validity were constrained. More recent work has examined the effects of expanding scope-of-practice laws for nurse practitioners, emphasizing access, utilization, and pricing outcomes, but typically without isolating the direct causal effect of provider assignment. The authors build on this literature by employing a quasi-experimental design that directly compares providers within the same institutional setting. The study also connects to broader work on human capital and productivity, including research documenting substantial variation in worker performance within occupations.

DataThe analysis uses administrative data from the Veterans Health Administration, the largest integrated health care system in the United States. The dataset includes approximately 1.1 million emergency department visits between 2017 and 2020, after nurse practitioners were granted full practice authority within the system. The data contain detailed information on patient demographics, comorbidities, prior health care utilization, and vital signs, as well as provider characteristics and treatment decisions. Outcomes include length of stay, treatment costs, hospital admissions, 30-day preventable hospitalizations, and mortality. The richness of the dataset allows for precise measurement of both inputs and outcomes, as well as extensive risk adjustment. The scale and administrative nature of the data enhance internal validity, though the institutional context—an integrated federal system serving veterans—may limit generalizability to other health care environments.

MethodsThe authors employ a quasi-experimental instrumental variables strategy to identify the causal effect of being treated by a nurse practitioner rather than a physician. The instrument is the number of nurse practitioners on duty in a given emergency department at a given time, which influences the probability of assignment to a nurse practitioner but is plausibly unrelated to patient characteristics conditional on fixed effects. The empirical model includes detailed controls for time and location, effectively comparing patients who arrive at the same facility under similar conditions but face different provider availability. This design addresses selection bias that would otherwise arise if nurse practitioners systematically treat lower-risk patients. The approach yields a local average treatment effect for patients whose provider assignment is influenced by staffing variation. Compared to standard multivariate regression, this method provides a more credible basis for causal inference, though it does not reach the level of experimental control offered by randomized controlled trials.

Findings/Size EffectsThe results indicate that nurse practitioners, on average, use more resources than physicians. Specifically, treatment by a nurse practitioner increases patient length of stay by approximately 11 percent and raises emergency department costs by about 7 percent. Nurse practitioners also increase the probability of 30-day preventable hospitalization by roughly 20 percent relative to the mean, suggesting lower average performance on this quality metric. No statistically significant differences are found in mortality. These average effects mask important heterogeneity. For less complex and lower-severity cases, the differences between nurse practitioners and physicians are smaller, while for more complex cases, the gap widens. Nurse practitioners are more likely to order diagnostic tests and consults, indicating differences in clinical decision-making. At the same time, the study documents substantial variation within professions. The distribution of productivity overlaps considerably, with nurse practitioners outperforming physicians in approximately 38 percent of pairwise comparisons. This finding implies that within-profession variation exceeds between-profession differences, complicating simple workforce substitution narratives.

ConclusionThe study provides evidence that, in emergency department settings, physicians exhibit higher average productivity than nurse practitioners when evaluated across both resource use and patient outcomes. However, the magnitude and direction of these differences depend on case complexity, and the substantial overlap in productivity distributions suggests that professional category alone is an imperfect proxy for performance. The findings highlight the importance of task allocation and the potential gains from matching provider skill to case complexity. From a methodological perspective, the use of a large administrative dataset combined with an instrumental variables design strengthens causal interpretation relative to much of the prior literature. Nonetheless, the absence of randomized assignment limits the scope of inference, and future research using experimental designs could further refine estimates of comparative productivity.