American medicine has belatedly entered the digital world with nearly universal utilization of the electronic medical record (EMR). Physicians have welcomed some aspects of the EMR and cursed others. As an industry, we were far behind most other commercial entities in “going digital,” and the versions of the EMR used in the United States at this time represent, in my opinion, only the earliest examples of an incomplete and less than totally useful digital product.
The Beast: In the United States, EMRs are designed to maximize billing collection and hence are filled with extraneous and irrelevant information. For example, both inpatient and outpatient clinical notes repeatedly contain the past history, the family history, a list of outpatient medications taken by the patient, and a long list of prior laboratory and imaging results. The result is a very long and less than clinically useful note.
What my colleagues and I usually do when perusing these electronic clinical notes is rapidly glance over the present illness, physical exam, and the few relevant laboratory or imaging results, working our way down as quickly as possible to the list of diagnoses, which contain important updated information concerning the patient’s progress, or lack thereof, and suggested evaluation and therapeutic plans. In an earlier era of hand-written notes, the SOAP (subjective, objective, assessment, and plan) format contained only brief summaries of the most recent and clinically useful information rather than an exhaustive compendium of largely unnecessary prior information that is currently copied and pasted into the EMR note. The EMR increases physician work time compared with the earlier era of hand-written medical records. In response to this increased burden, some physicians now dictate all their notes or employ scribes to do the paperwork; both techniques involve increased cost to the medical care system.
At this time, the use of the EMR has resulted in considerable physician dissatisfaction, depression, burnout, and early retirement.1 Unfortunately, current EMRs are still in a “childhood” phase of development and reflect, in my opinion, the old-fashioned and cumbersome method for calculating physician payment that is still in use in the United States. Hopefully, with time and the evolution of newer schemes for physician reimbursement, there will be concomitant improvement in EMR clinical notes that will more competently reflect clinical needs rather than billing requirements.
The Beauty: Despite these shortcomings, the EMR does have some major improvements over older systems of patient recordkeeping. Thus, when I open a patient’s EMR, I can see all of the recent clinical notes written by myself and by other clinicians. I can immediately access all recent laboratory tests and imaging studies and can also see a list of the patient’s current medications. Ordering further investigations and prescribing medication is also facilitated by the EMR once one knows the specific codes that are accepted by the computer program. My prediction for the future of the EMR is that once artificial intelligence is embedded in the software, this digital tool will become indispensable in daily clinical practice by assisting physicians with prescription renewals, drug interactions, patient adherence to prescribed medication, challenging diagnostic issues, and much more. I am told that it will not be many years before artificial intelligence is added to current EMR software.
Digital ECG interpretation (DEI) has been available for a much longer time compared with the EMR. Indeed, I participated in a study of 5 early versions of computer electrocardiogram (ECG) programs in the mid-1970s.2 In a recent issue of the The American Journal of Medicine,3 Professor Smulyan from the State University of New York at Syracuse described the various positive and negative features of DEI. My own experience matches that of Professor Smulyan. The beauty aspects of DEI include decreased time to interpret ECGs and highly accurate measurement of intervals such as the PR and QRS intervals. Measurement of the QT interval is less accurate and does require physician overread. The computer program reads ECGs in a manner different from the way humans perceive them. The computer performs multiple determinations of whether the ECG tracing is rising or falling, whereas humans use their excellent sense of pattern recognition. Because of this difference in approaches to interpretation, the computer reading is incorrect in approximately 20% of tracings (the beastly aspect). This is especially true for arrhythmia detection, which must always be confirmed by a physician experienced in electrocardiography. In the end, however, a highly accurate interpretation of the ECG is produced by the interaction of the computer program and an experienced physician. As noted above, DEI has been available since the 1970s and is currently a highly useful product when combined with physician overread. I, for one, hope that it will not take the same period of time for the clinical EMR to evolve into a similarly useful tool.
As always, I welcome responses to this editorial on our blog at amjmed.org.
To read this article in its entirety please visit our website.
-Joseph S. Alpert, MDa,b
This article originally appeared in the April issue of The American Journal of Medicine.
