New Hampshire Board of Pharmacy Proposes Rules Governing Artificial Intelligence in Pharmacy Practice

Newsletter

The New Hampshire Board of Pharmacy is considering proposed rules that would establish detailed standards governing the use of artificial intelligence and other technology in pharmacy practice. If adopted, the proposal could represent one of the first state pharmacy-specific regulatory frameworks to expressly address the use of artificial intelligence (AI) and clinical decision support tools by pharmacies.

The proposal appears in a broader rewrite of the Board's standards of practice rules under proposed Chapter Ph 500. Specifically, proposed Part Ph 522, Technology-Assisted Pharmacy Practice, would apply broadly to pharmacies, permit holders, and licensed pharmacy personnel using technology to support activities regulated by New Hampshire pharmacy law. The proposal expressly reaches technology used for prescription processing, patient communications, drug utilization review, adjudication, dispensing, verification, inventory management, quality assurance, compliance monitoring, and clinical decision support.

For pharmacies increasingly incorporating AI into clinical, operational, and administrative workflows, the proposed rules offer an early indication of how boards of pharmacy may approach the intersection of AI, professional judgment, and pharmacy licensure.

AI May Assist Pharmacy Practice, But Cannot Replace Professional Judgment

The central principle of the proposal is straightforward: technology may support pharmacy practice, but it may not replace the professional judgment, responsibility, or accountability of licensed pharmacy personnel.

Under proposed Ph 522.04, pharmacists would remain responsible for professional judgments required by law, including prospective drug utilization review, patient counseling, clinical assessment, final verification where required, and determining whether a prescription or medication order is appropriate to dispense. The rule explicitly prohibits the use of technology to automatically verify, approve, dispense, or suppress clinical information in a way that would eliminate legally required pharmacist review.

The proposal then applies that concept specifically to AI and clinical decision support tools. Under proposed Ph 522.09, AI and clinical decision support tools could not serve as the sole basis for decisions involving:

  • clinical appropriateness of a prescription or medication order;
  • prescription or medication-order verification;
  • dispensing a prescription drug;
  • overriding or suppressing a clinically significant alert;
  • making a therapeutic intervention;
  • making a determination requiring pharmacist judgment; or
  • denying, delaying, prioritizing, or otherwise affecting patient access to medication when patient safety or clinical appropriateness may be affected, unless the decision is reviewed by a pharmacist or other authorized pharmacy personnel acting within scope.

This is an important distinction. The proposal does not prohibit pharmacy use of AI. Rather, it seeks to establish a regulatory boundary between technology that informs professional decision-making and technology that effectively makes the professional decision itself.

Pharmacies Would Have Significant Pre-Implementation Obligations

The proposal also places substantial responsibilities on the pharmacy permit holder. Before implementing a technology-assisted workflow that materially affects prescription processing, dispensing, verification, patient communications, drug utilization review, inventory management, or clinical decision support, the pharmacy would be required to evaluate the technology's intended use, identify authorized users, determine whether the technology implicates pharmacist judgment or supervision, validate that the technology functions as intended, establish written policies and procedures, and train personnel regarding proper use, limitations, and escalation procedures. Documentation of these activities would need to be readily retrievable.

The AI-specific requirements go further. Before using an AI or clinical decision support tool, the pharmacy would need to evaluate and document:

  1. the tool's intended use;
  2. the workflow in which it will operate;
  3. the personnel authorized to use it;
  4. circumstances requiring pharmacist review or intervention;
  5. the process for accepting, rejecting, modifying, or overriding system-generated recommendations; and
  6. the process for monitoring performance, errors, bias, unsafe recommendations, and unintended consequences.

The express reference to bias is noteworthy. It suggests that a pharmacy's compliance obligations may extend beyond determining whether a tool technically works to evaluate whether its outputs create inappropriate or disparate clinical or operational outcomes.

Auditability Could Become a Key Compliance Requirement

One of the most consequential aspects of the proposal is its focus on records and auditability. Technology-assisted systems would generally be required to maintain records identifying the patient, prescription, medication order, drug, or transaction involved; the date and time of the activity; the individuals who performed, reviewed, approved, modified, or overrode the activity; any specific actions taken; and any pharmacist review or intervention required by law, rule, policy, or professional judgment.

The system would also need to capture system-generated recommendations, alerts, warnings, or interventions that were accepted, modified, overridden, or rejected. These records generally would have to be retained for at least four years and made available to the Board or the Office of Professional Licensure and Certification upon request.

For AI specifically, if a recommendation, alert, prioritization, or other output is accepted, rejected, modified, or overridden as part of a pharmacist's professional decision, the record would need to identify both the action taken and the individual responsible for the decision.

For pharmacies evaluating AI vendors, this requirement may be particularly significant. A tool that produces useful recommendations but does not create sufficient audit trails may not satisfy the proposed New Hampshire standard.

Vendor Contracts May Need Greater Scrutiny

The proposal also makes clear that outsourcing technology does not outsource regulatory responsibility. Under proposed Ph 522.13, the permit holder would remain responsible for compliance when using third-party technology. Pharmacies would need to ensure that vendor technology can support regulatory obligations concerning confidentiality, auditability, record retention, security, downtime procedures, and inspection by regulators. A vendor relationship also could not interfere with the Board's ability to inspect or obtain required records.

This provision could have practical implications for contracting with AI developers, pharmacy management system vendors, clinical decision support companies, and other technology providers. Pharmacies may increasingly need contractual rights regarding access to audit logs, data retention, system validation, cybersecurity, incident notification, regulatory cooperation, and changes to algorithms or system functionality.

AI Governance May Become Part of Pharmacy Quality Assurance

The proposal would also require pharmacies to periodically review technology-assisted workflows. Reviews must occur before implementation, following material changes, after certain errors or security events, and at least annually. Pharmacies would need to document those reviews and any corrective actions. The practical result is that AI governance would not be treated as a one-time implementation exercise. Instead, AI and other technology-assisted workflows could become part of the pharmacy's ongoing quality assurance and compliance infrastructure.

The Proposal Reaches Beyond Generative AI

Pharmacies should also take note of the breadth of the proposed definition of artificial intelligence. Proposed Ph 522 defines AI to include machine-based systems or software tools using algorithms, machine learning, natural language processing, predictive analytics, or similar functionality to generate outputs, recommendations, classifications, prioritizations, alerts, or other information supporting pharmacy operations or patient care.

Accordingly, these rules could potentially reach well beyond generative AI products such as conversational assistants. Depending on functionality, the framework could implicate tools used for drug utilization review, adherence prediction, prescription prioritization, clinical interventions, patient outreach, specialty-pharmacy workflow management, fraud or diversion detection, and other forms of algorithmic decision support.

That breadth is worth watching closely, as pharmacies may already use technology that falls within the proposed definition.

Practical Considerations for Pharmacy Operators

Although the New Hampshire proposal remains under development, pharmacies operating nationally should view it as more than a New Hampshire compliance issue. Boards of pharmacy frequently look to other jurisdictions when developing rules, and New Hampshire's proposal provides a potential template for pharmacy-specific AI oversight.

Pharmacy operators should consider beginning to inventory AI and clinical decision support tools currently embedded in their operations, including functionality supplied through existing pharmacy management systems and third-party vendors. Particular attention should be given to tools that influence dispensing, drug utilization review, patient prioritization, therapy recommendations, access to medications, or other activities involving pharmacist judgment.

Organizations may also want to evaluate whether their existing governance structures address several themes reflected in the proposal: defined intended use, human oversight, validation, documentation of overrides, bias and error monitoring, cybersecurity, vendor oversight, training, downtime procedures, and periodic quality review.

These concepts are increasingly emerging across state and federal AI policy and may become foundational expectations even in jurisdictions that have not yet adopted pharmacy-specific AI regulations.

What Next?

The New Hampshire Board of Pharmacy is continuing its work on the broader Ph 500 standards-of-practice rulemaking. The current proposal would establish Part Ph 522 as a comprehensive framework for technology-assisted pharmacy practice, including a dedicated section addressing AI and clinical decision support.

Pharmacies, technology vendors, and other stakeholders should monitor the rulemaking closely. Even if the final language changes, the proposal is notable because it moves AI regulation directly into the traditional Board of Pharmacy framework of professional judgment, permit-holder accountability, recordkeeping, quality assurance, and inspection.

Quarles' Health & Life Sciences Practice Group will continue monitoring developments involving artificial intelligence, pharmacy practice, and emerging state regulatory requirements. For questions regarding the New Hampshire proposal or the use of AI and other technology in pharmacy operations, please contact your Quarles attorney or:

Follow Quarles

Subscribe Media Contact
Back to Main Content

We use cookies to provide you with the best user experience on our website and to analyze statistics related to our website. To understand more about how we use cookies, or for instructions to change your preference and browser settings, please see our Privacy Notice. Please note that if you choose to reject cookies, doing so may impair some of our website's functionality.