Further Thoughts about the EHR Copy and Paste Issue

December 23, 2013

After the response – pro and con – to the crackdown on EHR copying and pasting, there is a true solution to capture both narrative and data from physicians.

The last article prompted a number of interesting comments ranging from "This is crazy," to "Right on." To expand the discussion, I will put on my naturalist's hat and develop a taxonomy (classification scheme.) There are two basic approaches that one can take to creating content and two basic ways to represent information that can address the dual need for narrative and data.

The two approaches are additive and subtractive. With the additive approach, you start with an essentially blank slate and add what you think (or are told) must be included. In the subtractive approach, you start with some chunk of content that will become the note whether or not you insert, update, and delete what is there. The two representations are data and narrative separate or data and narrative together.

The representation used by the typical EHR is data and narrative separate. Template-based systems are inherently subtractive in approach. The narrative consists of text (perhaps with pointers or references embedded in it) and the data consists of discrete elements (i.e. each data element is the answer to a single question.) Each widget that a user interacts with on a screen asks a question regardless of whether it's a text field, checkbox, or a dropdown.

Another representation with which you are all familiar is an HTML web page. HTML is one of a family of "markup languages" that are derived from pioneering work on SGML by Charles Goldfarb at IBM in the early 1980s. Markup allows text and data to reside in the same document and, depending on how that document is rendered (Web browsers render HTML in a specific way,) either the text, the data or both can be displayed or extracted. The following example from the Forbes website is illustrative. This is a portion of what is displayed:





CA is underlined indicating a link and that the system inserted the "CA +0.08%". This is proof that text cannot be reliably used as data. CA actually refers to California, not Computer Associates, but never mind that for the moment.

I call your attention to the HTML that generated the above:

<p>... At mid-day, he exercises, either playing tennis near where he lives in Pleasanton, <a href="/companies/ca/">CA</a> <span class="quotecard_hook initialized" data-ticker="CA" data-exchange="NASDAQ" data-type="organization" data-naturalid="fred/company/742" data-quotes-closing="32.515" data-quotes-now="32.48"><span class="wrapper decrease"><a href="/companies/ca/"><span class="ticker">CA</span> <span class="change">-0.11%</span></a></span></span>, a suburb of San Francisco...</p>

Look at all the data embedded in there: the ticker symbol, the exchange, yesterday's closing price, the current price, and a link to more info about Computer Associates. Not reflected here but important: Between the time I captured the webpage and the time I grabbed the HTML, CA dropped ~0.2%. This highlights the fact that sometimes there is merit is copying material and pasting it into a note. It documents the context in which decisions were made about the patient.

If this was a progress note, all the narrative and all the data would be contained in the same document - stored together, retrieved together, context intact (although wrong in this case.) There are myriad software tools that can parse these documents and extract either the data or the text. The trick is to build software that doctors will find "normal to use" that captures the appropriate, discrete data elements as they chart, generating both human-readable text and data and bundling it all into the encounter record. Voila! End of story.

In order to complete this thought I need to do something that I ordinarily eschew - to tell you that there is at least one EHR product that works this way. I know because I wrote it in 1993. It's called ChartWare. Since 1995, several thousand physicians and other practitioners have used it on and off. Some found preparing for EHR to be more than they could deal with. Some have been phenomenally successful, giving better care, increasing their productivity, lowering their overhead, and reducing the need for clerks. One doctor even wrote a song extolling it. ChartWare's laser focus on charting was ultimately its undoing; it's not certified and certification trumps utility.

The take home message is this: It is feasible to gather narrative and data without requiring physicians to perform duplicate work. Any data that you are willing to pay physicians to collect can be collected this way,  and  it will be more valuable because the context is preserved. It reduces the challenge of migrating to a new EHR. It's the right way to do charting in theory, it's the cost-effective way to do charting in practice and, lest I forget, it uses an additive approach. This means that nothing gets in there that the physician did not choose to include, even if what they include is a copy of the last lab results.

 Editor's Note: Daniel Essin serves as chairman of the board to ChartWare, but receives no financial compensation for that role.