The medical or health record serves as a comprehensive documentation of a patient’s medical history and care. Historically, these records were handwritten on paper, organized in folders by note type, and maintained as a single copy. The introduction of the Electronic Health Record (EHR) aimed to streamline the documentation process by making medical records legible and digitally accessible. Surprisingly, the transition from paper records to electronic versions has not alleviated the documentation workload and, as a result, healthcare providers continue to seek more efficient, time-saving methods, such as dictated recordings intended for transcription.
With all documentation methods, accuracy is crucial when documenting a patient’s encounter, as errors in the process can significantly impact the quality of care provided. Traditional methods, such as dictation by healthcare providers and transcription by external parties, have been used to streamline the creation of medical records. However, misheard words, misinterpreted phrases, and inaccurate transcriptions can lead to a flawed medical record, potentially resulting in mistaken diagnoses, inappropriate treatments, legal repercussions, and, as the case below demonstrates, irreparable patient harm.
Wrongful Death of a Patient Due to Medical Transcription Errors
On December 13, 2012, a Baldwin County Alabama jury delivered a $140 million wrongful death verdict against a hospital and its contracted medical transcription company for a woman’s death caused by a transcription error, which resulted in a fatal medication dosage.¹
A 59-year-old lifelong insulin-dependent patient with diabetes was discharged from the hospital to a rehabilitation facility after receiving treatment for a blocked dialysis access port. The physician’s dictated Discharge Summary was outsourced to a transcription company located outside of the United States. The transcriber mistakenly entered a dosage of 80 units of Levemir insulin instead of the intended 8 units. The unreviewed and unsigned summary was sent to the rehab facility as part of the patient’s admission notes and medication orders. As a result of receiving the incorrect dosage of 80 units of insulin, the patient experienced a cardiopulmonary arrest and sustained irreversible brain damage.
A misheard figure resulted in a typographical error that tragically caused a preventable medical mistake, leading to the patient’s unnecessary death.
Introduction of Artificial Intelligence and Documentation
Considering such preventable tragedies, can solutions—like artificial intelligence (AI)—enhance accuracy and reduce human error in clinical documentation?
AI offers opportunities to improve the efficiency and precision of medical record documentation. AI scribes and ambient AI tools are available for transcribing, summarizing, and integrating medical conversations into patient records. These technologies are designed to produce structured clinical notes, streamlining the documentation process for healthcare providers.
Unlike traditional dictation tools or human scribes, AI scribes function independently or with minimal human oversight, frequently integrating directly with the electronic medical record (EMR). Ambient AI refers to technologies that “listen” to patient-provider interactions, transcribe in real time, and generate clinical documentation.² These tools range from converting spoken dialogues into written transcripts to extracting relevant clinical information and creating concise clinical notes directly into the EMR for the provider to review. AI-driven digital scribes are able to capture the subtleties of medical discussions that go beyond the capabilities of human scribes or conventional dictation software, filtering out non-clinical conversations.
While AI has the potential to enhance efficiency and precision in documentation, there is a risk of preventable harm. AI-enabled technology may lead to a false sense of security, potentially producing misleading or inappropriate results. Even though AI-enhanced tools aim to reduce documentation time, it is essential for the practitioner to review and edit the notes generated by AI to ensure the accuracy of the medical record and prevent patient harm.
Guidelines for implementing AI-enabled documentation technology:
1. Establish an AI governance policy:
- Develop guidelines for the evaluation, implementation, oversight, and monitoring of AI to ensure that such technologies are used safely and effectively without compromising patient safety, privacy, and security, and to ensure alignment with existing laws and regulations governing the use of AI in healthcare organizations, including consent of all parties being recorded.³
2. Obtain informed consent from patients, either verbally or in writing:
Educate patients about the use of AI for dictation and transcription, including:
- Providing a statement saying that your practice's use of AI technology is intended to aid in creating thorough and accurate medical notes.
- How AI dictation functions and uses recorded information, i.e., transcribing the spoken word into a note.
- The duration of storage for recordings and measures in place to protect patient privacy.
- The benefits of improved accuracy and efficiency in documentation, including enhanced ability to focus on the patient during the encounter.
- The risks of AI, i.e., transcription errors, unable to recognize certain accents or speech patterns, or omission of non-medical conversations pertinent to the patient’s social history.
- An option to “opt-out” of AI scribe use and alternative documentation methods available.
3. Clarify the role of AI scribes:
- Reassure patients that AI scribes are meant to support healthcare providers, not replace them, to enhance documentation and data management without replacing care and treatment decision-making.
4. Ensure compliance with HIPAA regulations:
- Establish a contract with the AI vendor and include a Business Associates Agreement (BAA) to ensure that the patient data and information is protected.
5. Include disclosure in medical notes:
- Clearly state in the medical record that the note was created using AI technology and that you obtained the patient’s consent to use the technology i.e., “Additional history and documentation captured below via (electronic vendor).” or “Documentation services were performed after the patient or guardian consented to allow (electronic vendor) to record audio during this visit.”
6. Review AI-generated documentation prior to EHR integration:
- Physicians and other medical providers should review and edit the note prior to uploading it into the EMR.
While AI offers tremendous potential as an advanced tool to assist providers and healthcare staff, it is essential that human decision-making remains central to the care process. While discussions surrounding documentation will continue and evolve over time, the impact on patient safety must remain at the forefront.
Deborah Kichler, RN, MSHCA, is a Senior Risk Management & Patient Safety Specialist. Questions or comments related to this article should be directed to DKichler@CAPphysicians.com.
¹Cunningham Bounds, LLC. “Jury Holds Hospital & Transcription Company Responsible for Fatal Medication Error: $140 Million Verdict.” PR Newswire. December 17, 2012. https://www.prnewswire.com/news-releases/jury-holds-hospital--transcrip…. (Last reviewed January 29, 2025).
²Price, Lloyd. “AI Scribes and Ambient AI: Key Differences for Healthcare Providers.” Healthcare Digital. May 31, 2025. https://www.healthcare.digital/single-post/ai-scribes-and-ambient-ai-ke… (Last reviewed June 11, 2025).
³ECRI. Risks with AI-enabled health technologies. Hazard #1—2025 top 10 health technology hazards. Device Evaluation. December 3, 2024. https://members.ecri.org/guidance/risks-with-ai-enabled-health-technolo…
Additional resource:
Ervin, Yvette, JD. “The Role of Informed Consent in Medical AI: Balancing Innovative Advancements With Patient Rights.” CAPsules. January 16, 2025. https://www.capphysicians.com/articles/role-informed-consent-medical-ai…