Is Your EMR Charting Accurately?

For all the hope — not to mention time and money — being invested in EMRs as a way to improve health care, they’re still exquisitely prone to the age-old problem in IT: garbage in, garbage out.

Several writers have commented recently on whether you can believe what you read in an EMR. They raise serious questions as meaningful use Stage 2 draws near and providers’ care patterns become further enmeshed with their record systems.

One problem, wrote Dr. Rebecca Bechhold, a medical oncologist, is the information overload that an EMR can generate: “page after page of predetermined queries and stock answers that are repetitive and irrelevant after the first visit.” The truth, as in what’s really going on with the patient, might be in there somewhere, but she finds it hard to dig out.

Worse still, doctors sometimes just check “normal” for everything under the physical exam section because they’re in a hurry and entering the information is tedious, Bechhold wrote. Some pretty important history, such as an enlarged liver or an amputation, can be left out.

It might sound bad, but it’s human nature whenever there are too many boxes to check. However, for Bechhold, the key disadvantage isn’t a lack of facts, but of feelings.

“You cannot express the emotion and anxiety that is part of oncology care in a prepackaged document,” she wrote.

Software selection consultant Sheldon Needle, meanwhile, wrote about the pitfalls of taking an EMR prescription list at face value.

Take the patient who comes to the emergency room because of a car accident. If the patient’s regular doctor is linked with the hospital’s e-prescription system, a medication list might soon be forthcoming. But who’s to say there aren’t other medications in the picture, prescriptions written by a doctor who’s not tied in?

Needle’s advice: Ask a human, such as the patient or a relative.

“If something looks off on the electronic medical record,” he wrote, “question it.”’s own John Lynn, too, addressed the issue of trusting health care data, noting that doctors are receiving information from more sources than ever, including health information exchanges, patients and patient devices. It’s hard for physicians to know what’s reliable.

The obvious solution to trust issues seems to lie in user interface design. If the EMR is a good fit for the doctor’s workflow, the right data should end up in there.

Unfortunately, it’s not quite that simple. Bechhold noted that charts she receives from other doctors are sometimes configured to include every piece of data available for the patient, including all medications and test results.

The physicians, she wrote, want to be able to show that they reviewed all information if they’re ever sued.

Doctors and health IT companies have a way to go in understanding each other. Only then can there be full trust in EMRs.

About the author

James Ritchie

James Ritchie

James Ritchie is a freelance writer with a focus on health care. His experience includes eight years as a staff writer with the Cincinnati Business Courier, part of the American City Business Journals network. Twitter @HCwriterJames.


  • Approaches that help to reduce garbage in/garbage out include . . .

    1. identifying each posting at the EMR with a system generated date/timestamp and user “signature”.

    2. discourage the sharing of user logins (i.e. an assistant records data on behalf of a physician by logging in using the physicians user name/pass). Allow EMR date/timestamp recordings to indicate “recorded by ____ on behalf of ______)”.

    3. in respect of viewing e-charts, track all access to the EMR (i.e. assistant entered data “on behalf of” but physician later accessed and viewed that entry).

    4. include rule sets at forms that sniff out contradictory data/missing data during data recordings and issue warnings or, if warranted, a hard stop that requires a supervisor override.

    It’s difficult to prevent staff from copying prior session progress notes and pasting these as today’s session notes.

    As for ‘information overload”, it’s a good idea following a workflow step where, say, 10 pages of data has been entered, to add a narrative report generation step where the software system faithfully consolidates all of the data recorded to a narrative – the saving here is compression of sections on forms where no data was entered. This saves the physician from having to browse through the 10 pages when reviewing the e-chart.

    Nothing wrong either with giving an e-chart reviewer the ability to apply filters providing the filters do not survive the login session (otherwise the user might forget they had applied filters).

    Lastly, when distributing data to others, do this on a need-to-know basis (don’t send an entire CCD if the requestor only needs part of it).

  • Can’t say I’m familiar with commercial EMRs, but I would certainly hope there is a completely non-structured plain-text log entry, where the provider can record the session in plain English — independent of the other entries for the session, akin to what would have been done years ago, and quite readable.

    Not to say that the structured entries aren’t useful, but shouldn’t there be some place that’s human-readable?

  • “Take the patient who comes to the emergency room because of a car accident. If the patient’s regular doctor is linked with the hospital’s e-prescription system, a medication list might soon be forthcoming. But who’s to say there aren’t other medications in the picture, prescriptions written by a doctor who’s not tied in?”

    The obvious question, which I have asked for more than 20 years now, is how do know if, what and how much the patient is actually taking from the medications list?

    Question for you providers out there, would you rather have no information or partial information that is potentially/likely inaccurate?

    In other words, is no EHR better than some EHR?

  • The reality is that no matter how technically advanced our systems are, the human element is always necessary. Communicating with patients human to human and not relying solely on EMR is always going to be important.

  • @David, I would say that most newer EMRs recognize the need to accommodate recording of unstructured as well as structured data. The list of available forms should include a form that has nothing more on it but Patient ID, Name, Date, and a memo field.

    The ‘date’ should be the date the observation was made. The software system, in any case, will post the recording at today/now because clinicians don’t want to go to the e-chart, make a decision on what is there only later to find that someone was able to “insert” a backdated recording.

    If there is any likely confusion, three dates are needed a) session date (with the patient) b) the date the note was written up, c) the date the note was posted to the e-chart.

  • Does your EMR help compare and contrast your patient’s medicines from all sources with the medicines they were on when they last saw you? … So that you can instantly see what meds were increased, decreased, started, stopped or fell off the pharmacy’s list? And the reasons for the changes?

    The way I do it, the pharmacy’s list is highlighted green, imported, intermixed with my last visit’s list and alphabetized.
    So, if “digoxin 0.250 mg qd” is there in green highlight and also in white highlight, it hasn’t changed. It’s still there.
    If the green listing shows 0.125 mg, it was decreased by someone.
    If the white listing does not have a corresponding green one next to it, it’s either been dc’d or fallen from the pharmacy’s list.
    If there is a green one and no white listing for a med, it’s been added since patient was last seen.
    A keystroke removes superfluous lines.


  • Some great discussion here. Karl, thanks for posting those best practices. As to the unstructured section of an EMR, I think what Dr. Bechhold, quoted in the article, was saying was that even if the EMR allows for free-form comments, they’re often buried amid the checkbox-type info. Part of that wasn’t necessarily the fault of the EMR systems but of the doctors using them and choosing to send her details that she didn’t want.

  • As an alternative to getting lost in the structured data jungle of the EMR, being able to dictate a short and concise narrative that defines the patient-physician encounter is a much more intuitive approach. It is no longer necessary to dictate the entire report, only subsections that require a more precise narrative that tells the patient story. These “partial dictations” can now be converted to discrete data through Natural Language Processing and re-inserted into the subsections of the report. This prevents the physician from having to perform self-editing functions related to front-end voice recognition as well as an extra set of eyes and ears to ensure any inconsistencies are caught.

Click here to post a comment