While more New Zealand general practitioners are turning to AI scribes, University of Otago researchers note ethical, legal, and doctor-patient relationship challenges.
Recently, the university researchers surveyed primary care providers, including GPs, nurses, rural emergency practitioners, and practice managers, about their use of the AI-powered clinical documentation tool.
FINDINGS
The survey, conducted in the first quarter of 2024 and involving 197 participants, revealed that 40% used AI scribes to take patient notes. It also found that 66% read the terms and conditions of the tool, while 59% sought patient consent.
Eight in ten respondents found AI scribes helpful, with nearly half estimating it could save between 30 minutes and two hours per day; yet, others reported no time savings at all due to editing and corrections.
"Health providers retain professional and legal responsibility for the accuracy of their clinical notes regardless of whether they have used AI scribes. This requires checking all outputs for accuracy to avoid confabulations and omissions," the researchers noted.
The survey also uncovered major concerns, including compliance with New Zealand's legal and ethical frameworks, data security, errors or omissions, and data leaving the country. Some pointed out that AI scribes could not understand the indigenous te reo Māori language and New Zealand accents.
More than half, or 56%, of those surveyed, also pointed out having changed consultation dynamics with their patients when using the AI tool.
WHY IT MATTERS
The findings highlight the need to monitor and assess the impact of AI tools on clinical practice and patient interactions, said the study's lead, Angela Ballantyne, a bioethicist and professor at the University of Otago's Department of Primary Health Care and General Practice.
The professor also stressed that the benefits of using AI scribes must be balanced against patient rights and data security.
To maximise the benefits of AI scribes while reducing risks, according to the research team, several medicolegal issues need to be resolved:
- obligations around the retention and storage of health data
- how pausing or interrupting a transcription recording may affect accuracy and compliance
- ensuring patients can access and correct their health information
- risks posed when the original audio is deleted before transcripts are verified
- potential use of recordings or transcripts as evidence in legal proceedings
THE LARGER CONTEXT
Although it is slowing its uptake of digital technologies in healthcare, the New Zealand government still promotes the integration of AI to support healthcare delivery.
In July, an advisory group within Te Whatu Ora Health New Zealand formally endorsed two ambient AI scribe tools – Heidi Health and iMedX – for trial across public health services, following evaluation of privacy, security, and ethical standards.
Pharmac and Medsafe, meanwhile, have been exploring AI assistants to expedite medicine assessment and regulatory processes.
Contributing to the government's effort to establish the safe use of AI in healthcare, the Medical Council of New Zealand, the country's peak body for medical education and training, is expected to release guidance on AI use later this year.
