The surge in electronic health record adoption has led to a significant increase in patient-physician inbox messages, a trend further amplified by the COVID-19 pandemic.

This rise in digital communications has posed substantial challenges for health systems, prompting some to implement changes to manage the growing clinical burden.

A recent study published in JAMA Network Open attempted to shed light on the complexity of patient-physician electronic communications within a large integrated health system.

Conducted at Kaiser Permanente Northern California (KPNC), the study focused on understanding message content and variability, aiming to improve the efficiency of handling patient messages. The research team analyzed more than four million patient messages exchanged between April and August 2023, which were then processed through KPNC's Desktop Medicine Program, designed to streamline message routing and optimize physician workflows.

The report revealed more than three-quarters (77%) of messages received at least one label, with common categories including medications, skin conditions, messages with attachments and emergent content. Nearly a third of messages included multiple labels, indicating the complexity and diversity of patient inquiries.

The study also highlighted significant variability in message categories, with some topics showing consistent patterns, while others exhibited acute fluctuations. Controlled substances and medication-related messages, for example, demonstrated low variability, whereas topics like influenza vaccine and COVID-19 vaccine showed high variability.

The report noted one aspect of the Desktop Medicine Program was its real-time analysis of message content, which enabled the identification of emergent conditions and expedited regional review. This proactive approach helped shorten the time to clinical assessment for potentially urgent matters.

"Results suggest that a health system-wide approach to classifying patient messages paired with well-defined regional workflows can improve timely responses and substantially reduce physician inbox volume," the report noted.

Despite the program's success in resolving more than 1.5 million patient messages, the study acknowledged limitations, including its focus on a single health system and the exclusion of newer language models.

Researchers noted future enhancements are needed to address messages with multiple topics and improve content labeling and responses.

THE LARGER TREND

Health systems and care providers are looking at various solutions to better manage clinical documentation and EHR integration as they battle burnout and deal with staff shortages. Earlier this month, Rush partnered with Suki AI, a developer of a voice assistant for clinicians, to tackle physician burnout.

The health system is trialing Suki's platform across workflows in 30 specialties through bidirectional integration with Epic, aiming to achieve a 72% reduction in time spent on clinical documentation.

Meanwhile, Johns Hopkins is advancing AI in Epic chart summarization and has achieved success with AI-enabled patient portals, seeing promising results with ambient scribing.

Nathan Eddy is a healthcare and technology freelancer based in Berlin.
Email the writer: nathaneddy@gmail.com
Twitter: @dropdeaded209