09:06 AM
David M. Denton
David M. Denton
Connect Directly

Doctors Are Drowning In Data

Technology isn't enough to improve healthcare. Doctors must be able to distinguish between valuable data and information overload.

As a doctor, I know the value of information, but I also know what's worse than not enough information: misinformation or too much information. In this information age, we seem to have plenty of both.

No matter what you think or believe, you can find proof of it on the Internet. You can also find a million and one ways to decorate your living room, making it overwhelmingly impossible to decide which ideas to use. The Internet is great at quenching our attention deficits by providing novelty at every click. Indeed, we can spend hours reading, watching, listening, or commenting without accomplishing anything at all. On the other hand, we get access to excellent resources and minds, beyond what was possible in a non-connected world.

Modern medicine also struggles with managing information. In our lust for data, we have created systems that store every keystroke, scan, or import, in a limitless cloud. Discrimination is no longer necessary. The pertinent and the frivolous are stored side by side. We no longer have data; we have "big data." This allows the detection of trends and patterns that could never be identified with our smaller data sets. We are just beginning to understand its power.

[ Should be obvious: EHRs Must Solve Real Problems.]

Interestingly, however, while computers are great at sorting through data quickly and efficiently, humans aren't. In fact, "more," often clogs our ability to discern and decide. Additionally, computers can't distinguish good data from bad data. At present, humans are still required to use the data to make decisions and care for patients. Until we have computers that can form therapeutic alliances, be compassionate, diagnose conditions, and provide and coordinate reasonable treatments, we are still dependent on fallible biologic beings to provide our medical care.

One of the hopes of electronic health records (EHRs) is that they will revolutionize medicine by collecting information that can be used to improve how we provide care. Getting good data from EHRs can occur if good data is input. This doesn't always happen. To see patients; document encounters; enter smoking status; create coded problems lists; update medication lists; e-prescribe medications; order tests; find, open, and review multiple prior notes; schedule follow-up appointments; search for SNOWMED codes, search for ICD-9 codes, and find CPT codes to bill encounters (tasks previously delegated to a number of people); and compassionately interact with patients, providers have to take shortcuts.

(Source: Flickr user jfcherry)
(Source: Flickr user jfcherry)

To simplify the more cumbersome and involved process of documenting in EHRs, we use templates, checkboxes, and default reports. This standardizes the entry and ensures that all of the necessary bullet points are included. While this documentation allows more accurate CPT coding, it often doesn't reflect reality. Numerous patients with abnormal physical finding or other distinguishing features suddenly have normal exams except for the specific abnormality surrounding the chief complaint. Comatose patients are often "alert and oriented," and all ear infections look exactly the same -- "red and bulging." Template-based records are notorious for including things that were never done, such as performing a complete physical exam on a patient who came in with a splinter in a thumb. Or the record might detail a full review of systems -- including questions about exercise-induced chest pain and feelings of anxiety -- on a visit with a two-month-old.

Some systems limit the number of choices to describe things or use unnatural language or simple checkboxes to convey things previously communicated by narratives. While computers may be able to use this data, humans who have to use the information to decide what is going on have a harder time reforming the story from this cryptic information.

Additionally, ICD codes often lack the specificity required to communicate what is going on. Patients who once had 16p11.2 microdeletion syndrome suddenly have a diagnosis of 758.33 OTHER MICRODELETIONS. A patient born with an unbalanced AV septal defect, pulmonary atresia, ventricular inversion, and bilateral superior vena cava who underwent patch augmentation of the left PA and placement of a central shunt, followed by a bi-directional Glenn and finally a Fontan procedure is now synthesized down to 746.9 UNS ANOMALY HEART. (Unfortunately, ICD-10 isn't going to make this a lot better.)

While these codes can allow computers to quickly generate reports, they often simplify or group things together in ways that aren't useful to care providers. Much can be lost in the translation, making it impossible to appreciate the uniqueness or specificity of a problem needed to make

Next Page

good clinical decisions. Many systems don't allow any other details beyond the ICD-9 codes and terms in problem lists so that, for instance, in the case of UNS ANOMALY HEART, the provider knows nothing useful about the patient. In our rush to codify everything, we must not forget that uniqueness matters, and there is not a coding system designed that can surpass natural language at this time.

While misinformation creates distrust, too much information can conceal the pertinent. In a standard week, I receive charts from one or two new patients entering my practice. As a representative example, an electronic printout from a leading EHR vendor was faxed to my office last week (because the other practice and my group are on different EHRs that can't exchange information electronically). The chart consisted of 102 pages of information about a previously healthy five-year-old female who had been seen for healthcare maintenance and several sick visits during her short life. She had no chronic conditions or serious illnesses. It contained pages of anticipatory guidance, patient instructions, a long review of systems, release forms, timestamps, and generic advice about conditions. The only useful thing I could find in the printout was her immunization record, but I wasted 10 minutes discovering this.

Interestingly, in the same week, I received a faxed paper chart from a doctor who has been holding out against the trend toward digitization -- 10 pages of dictated notes and a summary page on a nine-year-old patient with asthma and a seizure disorder. Within seconds, I knew the medications he was on and had been on. I had an EEG and brain MRI report, and a good sense of his developmental status and asthma severity.

These are two extremes, and I certainly am not here to imply that paper is better than electronic data. I could have found two opposite examples to make the reverse point. The takeaway is that EHRs need better summaries and filters so that we can find and use the data we need. A few final examples illustrate this.

I often receive emergency department (ED) reports on my patients. As any provider in our community will tell you, these are onerous reports. Something as simple as an ear infection arrives as five to 10 pages of timestamps from every person who interacted with the patient in any way. Struggling through these reports, I can't always find out the diagnosis and treatment, but I always know which nurse, doctor, and lab technician saw the patient and the minute and second they did.

Finally, thanks to our health information exchange, I receive labs in my EHR every day from patients assigned to me at birth, while hospitalized, or while seen in the ED, who I have never seen before. There are often abnormal labs, so I have to decide what to do. Did someone else see and deal with these? Do I need to track down this patient and figure out what is going on? I spend no less than 30 minutes a day managing these issues. I also get copies of all labs and studies from patients I have cared for in the hospital. I have already seen and dealt with the labs, but have to look at each of them again, and go through the multiple-step process of signing them off, one by one, in our EHR. I am busier because of these things, but not a better doctor.

Information helps us learn. Information helps us make better decisions. Being able to get the right information helps us make the best decisions. The medical chart has become a source of information for insurance companies hoping to perform quick audits to minimize payments, for researchers hoping to capture data and perform studies, for patients interested in getting more information about their conditions, and for administrators hoping find new ways to cut costs.

With all of the new eyes looking at electronic health data, we must not forget that its primary purpose is to document useful information so that providers can provide high-quality care by having pertinent and correct information available. Finally, we must remember that each patient identification number represents a real human with his own beliefs, biases, struggles, and desires who is hard to represent by a few checkboxes or templates, especially if those templates aren't edited to reflect the truth.

Download Healthcare IT in the Obamacare Era, the InformationWeek Healthcare digital issue on the impact of new laws and regulations. Modern technology created the opportunity to restructure the healthcare industry around accountable care organizations, but IT priorities are also being driven by the shift.

David M. Denton is a board-certified pediatrician and member of the American Academy of Pediatrics. He is a partner of the Pocatello Children's Clinic in Pocatello, Idaho, and is affiliated with Portneuf Medical Center where he currently serves as the medical staff ... View Full Bio

We welcome your comments on this topic on our social media channels, or [contact us directly] with questions about the site.
Comment  | 
Email This  | 
Print  | 
More Insights
Copyright © 2020 UBM Electronics, A UBM company, All rights reserved. Privacy Policy | Terms of Service