TL;DR
Get workout gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
In a MedPage Today commentary published Oct. 20, 2026, nurse and cancer survivor Davis Austria describes how clinician-focused guidance left questions about treatment side effects unanswered during his care. He argues that health systems should test patient materials with patients for clarity, completeness and usefulness, especially as AI tools are used to rewrite them.
Davis Austria, an emergency and critical care nurse who has had head and neck cancer, argues in a MedPage Today commentary published Oct. 20, 2026, that health systems should test patient instructions with patients—not only for medical accuracy, but for whether people can understand and use them. His account comes as healthcare organizations adopt AI tools to write or rewrite patient materials, a practice he says could spread helpful explanations or repeat existing gaps at scale.
Austria writes that when he was diagnosed four years ago, his radiation oncologist showed him National Comprehensive Cancer Network (NCCN) treatment recommendations describing 10 chemotherapy rounds and 35 radiation sessions, then sent him home with the document. He calls the guideline accurate and evidence-based, but says it was written for clinicians and did not answer the questions he had as a patient.
His uncertainties included whether severe pain while swallowing was expected, whether his hearing would recover, and how he would learn to eat after a feeding tube was removed. The patient-facing NCCN guidance he later downloaded advised him to ask a healthcare provider about treatment for radiation burns. Austria says he used Aquaphor before radiation, then found another document that showed he had been using it at the wrong time.
He says he turned to a Facebook group for people with his cancer because the written materials and patient portal did not answer his specific questions. Austria now studies how health systems evaluate AI-generated patient information. He warns that AI can produce information that is wrong, incomplete or misleading, and says it does not replace clinicians. His central recommendation is to involve patients in evaluating materials before they are distributed.
Testing Instructions With Patients
The commentary focuses on a practical gap: information can be clinically accurate yet fail to tell patients what to expect or what to do. For people managing symptoms at home, that difference can shape how well they understand care instructions and when they seek help. Austria’s experience illustrates the risk without establishing how common it is across patients or health systems.
The issue gains urgency as AI tools are used to rewrite medical language. Such systems may make instructions easier to read, but Austria argues that readability alone is not enough. Health systems should test whether the material is complete, clear and useful for the people who will rely on it, across language, age and social circumstances. If the materials are flawed, automated distribution could reproduce the same problem widely.
Austria points to a randomized trial of 128 patients in Germany comparing conventional physician-written discharge summaries with AI-generated versions rewritten for patients. He reports that patients who received the patient-oriented versions scored 9.6 points higher on patient activation, a measure of preparedness to manage care. The commentary does not provide the trial’s full methods or establish that the result applies to every AI tool or patient population.
patient education materials testing kit
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Cancer Guidelines to AI Rewrites
Austria describes being treated both as a healthcare professional and as a patient. He cites a 2024 study in BMC Primary Care on doctors treating fellow doctors, which found that clinicians changed their communication and sometimes assumed professional patients already had relevant knowledge. He says his own clinical background did not prevent him from feeling lost after his diagnosis.
He also describes terminology in discharge paperwork that may be familiar to clinicians but unclear to patients: NPO means nothing by mouth, STAT means immediately, and PRN means as needed. As one example of a product response, Austria says Epic sells a patient-facing virtual assistant called Emmie, including a feature that rewrites after-visit summaries in plain language. The commentary does not evaluate how well that product works.
Austria’s views are his own, according to the disclosure attached to the piece; he says they do not represent the NIH, AIM-AHEAD, Ochsner Health or Xavier University of Louisiana.
“What I did not know was what was normal.”
— Davis Austria, in his MedPage Today commentary
How Well AI Materials Work
The commentary does not establish how widely patients encounter the problems Austria describes, or whether the AI tools now used by health systems produce materials that patients can reliably understand and act on. It also does not assess the performance of Epic’s Emmie assistant. The German trial cited by Austria involved 128 patients; the article does not provide enough detail to determine how broadly its results apply.
Austria’s personal account and the studies he cites support a call for patient testing, but they do not show that every patient-facing document is inadequate or that AI-written guidance is safer or more effective than material written by people. He says AI can be wrong or incomplete and stresses that it is not a substitute for a patient’s care team.
Questions Before Instructions Reach Patients
Austria urges health systems to ask whether their materials work for the patients they serve across differences in language, age and social circumstances; whether they have been checked for more than medical accuracy; and whether patients have tested their ability to understand and act on the instructions. He argues these questions should apply whether materials are written by staff or produced with AI.
The commentary does not announce a policy change, product launch or new study. The next step it proposes is practical: test patient instructions with actual patients before distributing them, and evaluate whether the information answers the questions people face while managing care at home.
Key Questions
What is the main development in this story?
Nurse and cancer patient Davis Austria published a commentary arguing that health systems should test patient-facing materials with patients for clarity, completeness and usefulness, particularly as they use AI to rewrite information.
What questions did Austria say his cancer materials left unanswered?
He wanted to know what symptoms were normal, whether his hearing would recover, and how he would eat again after his feeding tube was removed. He says the documents he received did not adequately answer those questions.
Does Austria say AI should replace clinicians?
No. He says AI can quickly assemble and explain information, but can also be wrong, incomplete or misleading, and is not a substitute for clinicians caring for a patient.
What does Austria recommend health systems do?
He recommends checking materials for more than medical accuracy and asking actual patients to test whether the instructions are clear, complete and usable before they reach a wider patient population.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
