Over 50% of Text in EHR Notes is Duplicate Clinical Documentation

More than half of all text in EHR notes is duplicated, according to a study published in JAMA Network Open that suggests the need for a new clinical documentation paradigm that supports the stability of shared information over time.

Researchers analyzed more than 100 million notes to characterize the prevalence of duplication in the EHR.

Duplicate content was prevalent in notes written by physicians at all levels of training, nurses, and therapists.

Physicians wrote the notes with the largest amounts of novel information and the longest notes. While comprehensive, the documentation was repetitive; physician EHR notes included 30 percent to 70 percent duplicate content.

The duplicated half of the content provides no new data and increases the time required for the reading clinician to discern which information is accurate.

“Overworked clinicians may be disincentivized from reading such a bloated record, missing valuable clinical context not easily found elsewhere, and leading to wasted time repeating past interventions or directly causing patient harm by missing findings requiring follow-up,” the researchers suggested.

Additionally, duplication creates viral copies of errors that can spread through a record until they are impossible to correct.

Given the high duplication rates, the study authors noted that this practice cannot be attributed to individual end-users behaving badly.

“Instead, our study suggests that duplication is a rational response of clinicians attempting to manage information in a documentation paradigm ill-suited to the task,” they wrote.

The prevalence of duplicate information suggests that a subset of clinical data remains true and relevant over time and should remain visible in the EHR.

“We propose that documentation systems be redesigned to take advantage of the stability of shared information over time,” the researchers said. “Duplication across authors might be addressed using collaborative documentation systems, in which each individual or team does not require a completely separate document.”

They suggested that stakeholders could address duplication across time by building documentation systems that enable editing and version history functionality to track changes to a single document.

“This functionality allows updates to be made without requiring creation of a new document, while maintaining old documents for medicolegal purposes,” the study authors wrote. “Multiple such systems have been described and implemented in medical contexts, described as dynamic documentation or the wiki model, but have not yet been widely implemented.”

Information scatter is another documentation hazard, the researchers noted. Data that should be stored in one place, such as the evolution of a chronic medical problem over years, is often stored across hundreds of separate notes.

“Under the current system, to keep relevant information about a patient in a single, up-to-date document, a note author will need to continuously create copies of old notes and add to them, rather than just editing old documents in place,” they explained.

“This practice directly contributes to textual duplication,” the researchers added. “Therefore, we cannot treat duplication in isolation, as unilateral restrictions on copy-paste behavior may exacerbate information scatter.”

The study authors emphasized that administrators should be wary of simple solutions such as bans on duplication, as this will only worsen other hazards without addressing the need to maintain data visibility, the authors wrote.

“In future work, we plan to perform qualitative examinations of duplicated text to further characterize its types and sources,” the researchers said.