In 2011, an article in The Journal of the American Board of Family Medicine commented that “information chaos” is routinely experienced by primary care physcians, and that this is “not just inconvenient, annoying and frustrating [but] it also has implications for physician performance and safety”. Much of the article remains relevant in today’s world.
Five primary factors were identified. Information scatter, overload, underload, conflict, and error. The following graphic from the paper containing a fictious progress note is extremely helpful in understanding the concepts.
The five categories can be defined as follows:
- Information overload occurs when there is simply too much data to “organize, synthesize, and draw conclusions from, or act on”. This is made worse by EMRs where even more data is available. A subset of overload is duplication where a record contains massive amounts of redundancy.
- Information underload occurs when there is insufficient data. It’s a common problem in all forms of practice that interoperability of systems is supposed to solve.
- Information scatter is when data is spread throughout the set of records. To cope with scatter the physician needs to read the entire set of records and attempt to put the jigsaw together.
- Information conflict occurs when records contradict each other.
- Errorneous information occurs when the patient, or the person entering data into the chart, gives the wrong information. Once this data is on the chart it can be remarkably hard to remove.
While the paper by Beasely et al is focused on primary care physicians and their experience with EMR, the findings are broadly applicable. The eHealth Technologies referral pathways suite contains a records aggregation suite that addresses each of these problems.
First, let’s discuss overload. A typical oncology case may contain hundreds of pages of medical records drawn from a wide variety of providers. You can expect to see flow sheets, history and physicals, operative notes, procedure notes, and on, and on. There is simply no way that a physician can read all this material, and even if they could I’m not sure it would make much sense. However, by collating the data, sorting it, making the most relevant data available first, highlighting key terms and adding clinical summarization the problem becomes more tractable.
Underload is more complex. It’s common in our business to have patient’s who don’t remember where they have been. Sure, you can look some of this data up in insurance records and you can tease some of it out of other provider records, but ultimately a human being is needed to do the detective work. If the patient says they went to the clinic near Joe’s pizzeria then google street view becomes a useful tool.
Scatter can be handled using the same approaches as overload. Summarization is key here.
Conflict and erroneous information are difficult issues that don’t lend themselves to systems yet. For unstructured data the best approach is a clinical person reading and correcting the data. It’s my hope in the future that we’ll be able to “score” or “annotate” data in EMR’s to reflect accuracy. Ideally the patient would participate in this process!
In my experience the five sources of information chaos are real. Will, EHR’s fix the problem? Certainly not today, but with careful design the future is hopeful. Beasly et al are hopeful.
EHRs contribute to more timely and available information but caveats exist, many of which have been discussed.11,45 EHRs are generally designed to facilitate data entry to conduct and document the process of care. As more and more data are available in an EHR, there is an even greater need for improved search methods and display techniques to present the data needed at the time of the patient visit. Ideal EHR design would allow relevant, needed information to be pushed to the clinician based on the reasons for the visit.