PRACTICE PEARLS

To improve our clinic’s performance on preventive care metrics, we implemented a protocol to discuss preventive care at acute care visits. To facilitate this, our EHR displays a “Best Practice Alert” when preventive care gaps exist in the patient’s record. This prompts rooming staff to use the following script: “In addition to (clinician) seeing you for (acute concern), it appears you have some preventive care due, specifically (preventive services). Would you like to schedule a separate appointment to discuss the preventive care, have (clinician) discuss it at this visit, or discuss this at another time?”

If patients opt to discuss it that day, rooming staff click on the EHR alert. With one click, care gaps become pending orders and associated diagnoses are added to the visit. Clinicians then add the preventive care discussion to the visit agenda. After the first six months, this system produced significant gains in colon and breast cancer screening, winning over even our clinic’s skeptics.

Chatbots such as ChatGPT and Google BARD can play a role in medical communication, but exercise caution with this technology because of the risk of misleading or incorrect information. We conducted an experiment last year, feeding ChatGPT a series of questions about common health conditions and then assessing its accuracy and coherency. For example, we asked what percentage of people experience nausea with a concussion?

ChatGPT’s response: “The prevalence of nausea in people who have a concussion varies depending on the study, but it is a common symptom. According to one study, up to 40% of people with a concussion experience nausea...”

We asked for more information about the 40% figure.

ChatGPT’s response: “… the study I was referring to is ‘Acute Symptoms and Time to Recovery Affer Sport-Related Concussion in Children and Adolescents’ by William P. Meehan III, Rebekah C. Mannix, and R. Dawn Comstock. It was published in the journal JAMA Pediatrics in 2013.”