EHRs are overflowing with copy-and-paste records, JAMA study shows
For all the benefits brought about by electronic health records, it's long been known that they have their pitfalls, whether it's ungainly user experience or agita caused by alert fatigue.
Another major risk for EHRs is the temptation toward "note bloat" caused by caregivers' easy ability to copy-and-paste data from other parts of the chart. This defeats the purpose of electronic documentation, of course, creating a large and unwieldy record that can be hard to make sense of – potentially putting patient safety at risk.
A new study from the Journal of the American Medical Association doesn't offer much in the way of encouragement.
Researchers at University of California San Francisco took a look at thousands of progress notes, written by nearly 500 clinicians over eight months in UCSF Medical Center's inpatient Epic EHR.
They found that only a small minority of them were manually entered – but more than 80 percent of the notes were imported or copied from elsewhere.
Other studies on copy-and-paste have been limited in their ability to quantify just where and when EHR text originated. But a recent software update to Epic allowed the UCSF team to examine the charts with "character-level granularity," researchers said.
"The EHR now identifies the provenance of every character that is present in a signed note – that is, whether the character was typed fresh ('manually entered'), pulled from another source such as a medication list ('imported'), or pasted from a previous note or elsewhere ('copied')," according to the JAMA report. "Clinicians can opt to see this information, which is hidden by default but is logged in the EHR for every note written since the upgrade."
By examining the proportion of manually entered, imported and copied characters in the notes, UCSF researchers were able to learn some revealing things about documentation practices – and the people doing the documenting.
The team analyzed 23,630 inpatient progress notes written by 460 caregivers who were either direct care hospitalists, residents and medical students.
For the study as a whole, the UCSF researchers found that 46 percent of notes were copied and 36 percent were imported. Just 18 percent of the text was entered manually.
Only 12 percent of Residents entered text manually while 51 percent copied. For medical students, more 16.2 percent entered manually and 49 percent copied while 14 percent of hospitalists opted for manual entry versus 47.9 percent who copied.
Meanwhile, hospitalists wrote the shortest notes (5006 total characters), compared with residents (6720) and medical students (7053).
"The traditional goal of progress notes is to provide a concise, up-to-date reflection of the patient’s condition and the clinician’s thought process," said UCSF researchers. "However, copying or importing text increases the risk of including outdated, inaccurate, or unnecessary information, which can undermine the utility of notes and lead to a clinical error."
They noted that the research could prove enlightening for vendors – hoping that it could "spur EHR design that makes copied and imported information readily visible to clinicians as they are writing a note but, ultimately, does not store that information in the note."
Hospital CIOs also have a role to play when it comes to documentation improvement. Geisinger Health System previously offered its staff (clinicians and coders alike) an array of tools to help with documentation, from speech recognition to auto-generated text.
But all those ancillary tools were also leading to a lot of extraneous and unnecessary data. So Geisinger developed guidelines to help streamline clinical documentation, encouraging caregivers to write quality notes.
As part of that effort, it used analytics to assess the items in patient problem lists and then develop single, standardized list across the organization, with the most common medical complaints given pride of place at the top, said Geisinger CIO Frank Richards, who also noted that the health system was looking natural language processing as a means of "extracting more value from dictation."
To help force the change, Geisinger changed its policies on copy-and-paste – prohibiting any free text from being replicated in the EHR without noting the original author, date and time of entry. Clinicians were also barred from pasting images from Geisinger's PACS into the EHR.