Author Identifier (ORCID)
Abstract
Background: Health-professional mobility agreements (eg, compacts, mutual recognition arrangements, memoranda of understanding) are increasingly used to address workforce shortages, support crossborder practice, and enable new technology-mediated modes of service delivery. Regulators need rapid, reproducible ways to assess whether such instruments protect the public while facilitating proportionate professional mobility.
Purpose: To develop and internally test a 10-dimension, 41-subdimension assessment framework for health-professional mobility agreements; examine whether large language model (LLM)–assisted coding can approximate expert legal regulatory assessment; and derive an empirical taxonomy of mobility and other similar instruments.
Methods: A focused documentary synthesis of legal instruments, academic literature, and intergovernmental guidance was conducted to inform framework development. Two expert assessors independently coded a purposive corpus of 70 instruments enacted or issued between 1902 and 2025. Claude Sonnet 4.5 (Anthropic) was used as an LLM-assisted third coder using a structured prompt, data dictionary, and audit trail. Reliability was assessed using Cohen weighted kappa and Krippendorff alpha. Document-grounded category development, network analysis, and nonparametric statistics were used to examine category stability and performance patterns.
Results: Pilot human reliability was substantial (weighted κ = 0.782), and full-sample human agreement was high (weighted κ = 0.853). Agreement across two human coders and the LLM was excellent in the full sample (Krippendorff α = 0.900). Nine instrument categories emerged: six purpose-aligned professional mobility categories and three purpose-mismatched migration, trade, or labor-protection categories. Interstate compacts formed a clearly distinct high-performing category (mean 2.30/3.0), followed descriptively by regional integration directives (1.68), domestic regulation (1.13), bilateral recognition treaties (0.96), multilateral recognition arrangements (0.94), and bilateral recognition memoranda of understanding (0.77). Purpose-mismatched instruments scored lower on the mobility rubric, reflecting misalignment between their objectives and the assessment lens rather than necessarily poor design for their own purposes. LLM-assisted coding reduced initial coding time from approximately 6-8 hours to 7–10 minutes per instrument, but this process required human oversight to address incomplete parsing, overinterpretation, and clause-citation risk.
Conclusion: The developed framework provides an internally reliable and policy-relevant basis for comparative assessment of health-professional mobility instruments. The findings suggest that mobility performance is driven less by legal label than by embedded governance architecture, operational infrastructure, enforcement, data systems, and purpose alignment. External, multilingual, and regionally expanded validation is required before the framework can be used as a global mobility-governance index or routine regulatory decision-support tool.
Keywords
health-professional regulation, interstate compacts, large language models, mutual recognition, professional mobility, WHO global code of practice
Document Type
Journal Article
Date of Publication
1-1-2026
E-ISSN
21558264
ISSN
21558256
Publication Title
Journal of Nursing Regulation
Publisher
Elsevier
School
School of Nursing and Midwifery
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Benton, D. C., & Benton, A. D. (2026). Governing professional mobility: An LLM-assisted framework and empirical taxonomy of health-professional mobility agreements. Journal of Nursing Regulation, 17(3), 298–307. https://doi.org/10.1016/j.jnr.2026.07.003